Step 1 : The Line stage

stephie's place

Moderator: sbench

Post Reply
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Step 1 : The Line stage

Post by sbench »

Intent is to allow folks the opportunity to listen to this in their own comfortable environment. Line / control (shown below ... but there's no "guts" in it yet, just switches, although the power supply operates)
has the following intended functions
1. Level control (top left) Transformer attenuator
2. Balance (bottom left) 11 posn switch. Method TBD
3. Selector, 4 position (top center)... e.g., CD, Tun, Aux, Pho
4. Volume/Loudness switch 4 position (bottom left middle)
5. Record Monitor switch 3 posn (bottom right middle) .. eg. norm/CDR/Tape
6. Treble, 11 position (top right)
7. Bass, 11 position (bottom right)

Yep, that's a Mullard EL34 sitting in a socket just to the left, shown for size.

Initial buildup will have source stuff, the volume/loudness controls, D3a/7721 amplifier stage and transformer (which would feed the tone controls).

When built up, we'll ship this critter around for anyone who wants to to listen.

Steve
Attachments
pre.gif
pre.gif (217.65 KiB) Viewed 27934 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Re: Step 1 : The Line stage

Post by dave slagle »

sbench wrote: 2. Balance (bottom left) 11 posn switch. Method TBD
this is the nut i am interested in cracking. (balance control for a stereo 3dB/step autoformer.
Initial buildup will have source stuff, the volume/loudness controls, D3a/7721 amplifier stage and transformer (which would feed the tone controls).
i guess this is where the 3:1 fits into the picture. Attached below are the frequency sweeps of a first go. The design fits the needs put fourth for DCR, current and inductance. One plot represents the worst case situation which is the signal generator (50r) driving the primary and the load on the secondary being the scope probes and the input to the meter. the other plot represents a guess at the source and load impedances. given the actual use will toss all kinds of different loads at the device, i am somewhat pleased that nothing squirrelly happens below 100K in either situation. fwiw the behavior around 200K still makes me a bit "worried"
Attachments
bench 3-1.gif
bench 3-1.gif (10.35 KiB) Viewed 27930 times
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Very nice. The >100kHz stuff may be controllable in other ways too. You're right, there's an "odd load" on that transformer. It'll be interesting to see what that contributes...

Audio path expected is...

source --> source selector --> "tape" monitor selector --> inductive volume control --> D3a stage --> 3:1 transformer --> tone controls --> power amp --> speakers

there is expected to be pickoff for
recorder source directly after source selector
alternative pa1 after 3:1 transformer
alternative pa2 after tone controls

Steve
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

3:1 version II

Post by dave slagle »

I did a revision of the 3:1 pictured above trying to deal a bit with the 200K peak. I was able to push it out to around 300K, but this came at the cost of dropping the -1dB point from around 120K to 90K. I cooked my AC meter to do the actual measurement so alas no plot.

Essentially what i did was increased the interwinding spacing to decrease capacitance at the cost of more leakage. the results seem to be inline with what would be expected, since the peak moved up in frequency (less interwinding C) and the earlier rolloff is what is expected with more leakage.

as a plus to this design, there is a third winding with a 40:1 turns ratio "buried" within the rest and it measures much better than the previous version where it was tacked onto the end.

i have one more trick to try with the 3:1 for curiosity sake and to possibly find out something more about the 40:1, then we should be ready to go!

dave
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Very nice! It'll be interesting to see and hear how it goes. |'ll be sure to measure (and report) the HF behavior.

There's enough built up in the physical box that I can add the line stage reasonably easily. Also, with I-O ports available, this will give us all a quick box to ship around and listen to.

Envision the loudness/volume/attenuator, line gain stage/xfmr to provide "critical mass" to start some listening tests with.

Steve
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

i used my scope to plot the FR of the 2nd version... it was actually better than i had previously thought.

i actually bought some top end in addition to pushing the resonance beyond 200K

dave
Attachments
bench2.gif
bench2.gif (13.53 KiB) Viewed 27874 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Rev. 3 is done

Post by dave slagle »

OK...

here are the plots for the three versions of the 3:1's that will double as the plate choke with a 40:1 ratio cathode winding with DCR's to match whats needed for the power amp.

there is an unloaded and loaded plot for each and the loading is the same for each case. I am not a fan of tweeking a trannie for great test results and steve will provide the only measurement that matters :-)

here is the quick case history.

all three have the same number of turns and the same sectioning, the only difference is the interwinding spacing and the way the wire was put down for each section.

#1 represents a method to place the wire that i have become fond of
#2 uses the same method with more insulation between layers
#3 uses the same insulation as #2 but with a more traditional lay to the wire.
Attachments
bench3.gif
bench3.gif (24.59 KiB) Viewed 27856 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Rev. 3 is done

Post by dave slagle »

OK...

here are the plots for the three versions of the 3:1's that will double as the plate choke with a 40:1 ratio cathode winding with DCR's to match whats needed for the power amp.

there is an unloaded and loaded plot for each and the loading is the same for each case. I am not a fan of tweeking a trannie for great test results and steve will provide the only measurement that matters :-)

here is the quick case history.

all three have the same number of turns and the same sectioning, the only difference is the interwinding spacing and the way the wire was put down for each section.

#1 represents a method to place the wire that i have become fond of
#2 uses the same method with more insulation between layers
#3 uses the same insulation as #2 but with a more traditional lay to the wire.
Attachments
note that the frequencies represented are only from 100K to 1 mhz.
note that the frequencies represented are only from 100K to 1 mhz.
bench3.gif (24.59 KiB) Viewed 27856 times
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

#1 represents a method to place the wire that i have become fond of
#2 uses the same method with more insulation between layers
#3 uses the same insulation as #2 but with a more traditional lay to the wire.
Verrry interesting. What it looks like to me is #1 is more "controlled" in that the resonance is somehow smoother. That is, the rate of change of response overall is lower in #1. That usually translates to better sound. However, generalizations *can* somehow be easily broken.

For some reason that I can't articulate, #3 disturbs me the most. I'm gonna have to stare at that plot for a while to articulate the answer, but from from what I see, the preference is #1, then #2, then #3.

Steve
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

#1 and #3 both bother me because the second resonance is greater than the first one... not sure what that means, but it bothers me.

The smoother behavior of #1 might very well be due to a lack of sample points.

I like #2 because it seems the most predictable and seems like it "rings out" naturally. To me the size of the peaks just shows the Q of the resonance. From my autoformer experience, a large peak beyond 200K was less objectionable than several smaller peaks at a lower frequency.

i might give a shot at #4 to see if we can come to a good compromise.

dave
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

I like #2 because it seems the most predictable and seems like it "rings out" naturally. To me the size of the peaks just shows the Q of the resonance. From my autoformer experience, a large peak beyond 200K was less objectionable than several smaller peaks at a lower frequency.

i might give a shot at #4 to see if we can come to a good compromise.
I tend to agree. Once you're out that far, the effects tend to be inaudible, so long as the amp itself behaves (IM issue).

Say. I've got some great mono recordings. Howzabout sending one of each and I'll listen mono to see what it sounds like.

Steve
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

sbench wrote: Say. I've got some great mono recordings. Howzabout sending one of each and I'll listen mono to see what it sounds like.
thats the answer i was looking for :-)

i'll send a care package with a pair of cores and an assortment of bobbins and you can have at it. Bobbin #1 is a bit ugly, but as clip leads go, it should do fine. I'll try to get to bobbin #4 tonight to see if i can mix it up a bit.

dave
Raj Gupta
Posts: 37
Joined: Fri Feb 11, 2005 2:19 am

Post by Raj Gupta »

Steve:

What do you generally prefer to do with mono LPs? Do you take the two stereo cartridge feeds and sum them, take one feed and wye it to both channels, listen to one speaker by itself?

-j
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

What do you generally prefer to do with mono LPs? Do you take the two stereo cartridge feeds and sum them, take one feed and wye it to both channels, listen to one speaker by itself?
Hi j and Dave,
Varies somewhat. Usually, I simply ignore the fact that the source is mono or stereo, so L goes to L amp and L spkr, and R goes to R amp and R spkr, which used to be called "binaural".

Sometimes, if the source is really dirty, usually adding (paralleling) the channels eliminates some noise.

If I'm in strict mono phase, I'll add and play thru one amp/one spkr set.

For these tests, what I'll probably do is add the source, play thru one line amp and try both possibilities... play thru 1 poweramp/spkr then repeat thru both power amps/speakers. For mono, I prefer using the 841 A2 amp (did you hear that one?)
i'll send a care package with a pair of cores and an assortment of bobbins and you can have at it. Bobbin #1 is a bit ugly, but as clip leads go, it should do fine. I'll try to get to bobbin #4 tonight to see if i can mix it up a bit.
Great. That oughta work just fine. It'll be interesting to see how this thing starts to develop.

Steve
Raj Gupta
Posts: 37
Joined: Fri Feb 11, 2005 2:19 am

Post by Raj Gupta »

For mono, I prefer using the 841 A2 amp (did you hear that one?)
I think I heard the Original Flavor only, with the extremely inexpensive tubes.

It's clear to me that summing the channels of a mono source sounds better than leaving them separate. But some people like to have the effect where the music comes from the center and any noise is dispersed in space around the outside. I find that more annoying than having the noise partially cancelled, and more "masked" coming from "behind" the central music image.

But WTFDIK? If I were king, preamps would have a stereo blend control instead of balance control. Mostly just for those first couple Beatles records, and for summing monos, but there are occasionally other times when it'd be nice to narrow things down a little.

-j
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

I think I heard the Original Flavor only, with the extremely inexpensive tubes.
Ohhh. Must nota heard that one then, as there's only 1 flavor... 8532 differential driving 6H30pi driving 841. Direct coupled, no matrix. The 841 looks exactly like an 801 except it has finer pitch grid spiral for higher mu.
It's clear to me that summing the channels of a mono source sounds better than leaving them separate.

...

But WTFDIK? If I were king, preamps would have a stereo blend control instead of balance control. Mostly just for those first couple Beatles records, and for summing monos, but there are occasionally other times when it'd be nice to narrow things down a little.
Quite interesting. My "oddball" matrix amp with the YL1060 (which has been on loan for over 2 years now) has blend/antiblend control on it. Interesting effect.

Say, maybe I oughta put a 6th position on my balance control... the 6th position automagically sums the signals together. Dave's putting a 3rd winding on the Line amp transformer, which I can use for balance or, if I really think about it, use it for blended channels. Hmmm.

Steve
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

i'll send a care package with a pair of cores and an assortment of bobbins and you can have at it. Bobbin #1 is a bit ugly, but as clip leads go, it should do fine.
Care package received. I'll start assembling things . This looks like it ought to be interesting.

Steve
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

***updated***
---> Listening tests updated.
---> Data presented for all 4 transformers.
---> Balance discussion expanded

Well, did a bunch of work on this thing. Got the attenuators inserted into the system, built and debugged the D3a/7721 line stage. Ended up with diode biasing with additional current thru the diodes to lower the Z and to linearize things. Overall, with the 3:1 stepdown transformers in place, I end up with 28dB gain, which is more than enough to allow for the 20dB loss for tone controls later. Might try using the tertiary winding in the cathode at some point for a no R as well as no C linestage :-) Running about 15-18mA thru the transformer.

Have listened to all 3 'formers: #2, #3, #4. Each has its own characteristics and good and bad points. I'd strongly suspect you'd get proponents for each one of these, so let me describe what I'm hearing perhaps... #4 is the "in your face" upfront, very dynamic transformer. Perhaps too dynamic, but many people prefer that. #3 is the "laid back" transformer. Not as dynamic, but relaxed, competent and -well- sweet. #2 is THE transformer for rock, but a little "rough" on other source material. I could probably live with any of them, but I'm leaning towards #4 at this instant. Christian (Rintelen) sent me a CD of "Bimini" last year made from vinyl and my LR phono preamp. Have mixed that into mono and listening as I type this... wonderful. #4 definitely tops on that cut.

Hey, whadjya know... that tertiary winding works wonderfully as a "balance" winding. Get 0.95dB "gain" with that winding aiding, and 1.2dB "loss" with that winding opposing the secondary winding. The effect is subtle but quite satisfactory overall using the circuit shown in the "balance" thread. For transformer 2 and 3, the balance "only" works up to 50kHz or so. For transformer #4, however Dave wound that one, the tertiary maintains "control" up to 200kHz, and the balance follows the overall channel response. (see the graphs below).

Dragged out the ol' Ballantine linear dB wideband VM to do the frequency response measurements. I still like that particular meter. (320A) In the data below, lowest freq is 10Hz and above 100kHz, the frequencies used were 100k, 110k, 120k, 130k, 140k, 150k, 160k, 170k, 180k, 190k, 200k, 210k, 220k, 230k, 240k, 250k, 260k, 270k, 280k, 290k, 300k, 350k, 400k, 450k, 500k, 600k, 700k.

Attached are frequency response measurements for #2, #3 and #4. Note that "0 dB" is about 28dB gain overall. Measurements with 600 ohm source Z thru "0 dB" attenuation (straight in), but including all the wiring and line stage stuff. I ended up with the "gold" spacer on the transformers as this seemed to produce the highest measured L at about 20mA DC and I'm running slightly lower than that in the line stage.

Also I've included response associated with using the tertiary winding for balance.

Next thing on the agenda I think I'll try that "no R" stage, using one of the tertiary windings in the cathode, just for the halibut.

Steve
Attachments
la_bal_4.gif
la_bal_4.gif (14.44 KiB) Viewed 26176 times
la_fr_b.gif
la_fr_b.gif (18.53 KiB) Viewed 26176 times
la_bal_3.gif
la_bal_3.gif (14.16 KiB) Viewed 26185 times
la_bal_2.gif
la_bal_2.gif (14.91 KiB) Viewed 26185 times
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

OK. Here's the update on the transformer for the line stage as well as a few changes overall I did on the system.

Transformer: Dave, you're right. Attempting to do too many things (your universal transformer) is not the right approach. Did not get to where I wanted to go using the tertiary winding in the cathode. (Coupling issues I think - That makes the input stage D3a to EL34 part essentially "one" winding with the cathode winding the second winding.) The transformer for the line stage can remain pretty much like #4 (3:1:T). The "balance" tertiary works perfectly. What we can do is wind that few turn tertiary with heavier wire (yay sez Dave) as there's no resistance "spec" and the lower resistance keeps the Zout low. The inductance seems to be OK Dave, 'cause I changed the line stage... see below. Could I have a couple more bobbins thusly wound? Should be "finals".

Line Stage... Interesting issue with D3a/7721. At the bias required, there was too much grid current happening, so that the attenuator was indeed "noticing" it, particularly in loudness mode. (bias point would change slightly at different attenuation/loudness settings, and when that happens, POP. Thus the "unpleasantness"). Also, the gain was too high. That combo made the system unpleasant. Changed to 5687 running 15ish mA at about 7.5V bias (from a 7.5v 5 watt zener. This had an effective measured dynamic resistance at 15mA of slightly less than an ohm!) Made all the difference in the world. With the lower rp of the 5687, the inductance is just fine. In fact, the way the transformers are wound, I may end up putting the halves of the secondary in series if I want more gain.

Attenuator... the one limitation I found I noted above. With tubes that require relatively low bias voltages, the bias point changes from volume/loudness etc, causing a pop. Also, the grid current prevents a "hitless" switch, as either make-before-break or break before make switches have issues (either changing tube bias or shorting the transformer monentarily). With high bias (5687 at 7.5V qualifies, as would 76, essentially all DHTs etc) tubes, the grid current is low and constant so there is no issue. Also found the overall loudness effect less subtle on the "final" versions. Ended up changing the interleaving to 2:1 which works just fine.

Hey Dave, whatchya wanta tackle next? The tone controls?

Steve
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

sbench wrote: Did not get to where I wanted to go using the tertiary winding in the cathode. (Coupling issues I think - That makes the input stage D3a to EL34 part essentially "one" winding with the cathode winding the second winding.)
i'm not convinced that given the required DCR's for that situation the coupling can be made much better. I do think trying to make it happen with all the other requirements taught us something :-) If we wouldn't have tried and failed on that aspect, the balance issue would never have been realized.
The transformer for the line stage can remain pretty much like #4 (3:1:T). The "balance" tertiary works perfectly. What we can do is wind that few turn tertiary with heavier wire (yay sez Dave) as there's no resistance "spec" and the lower resistance keeps the Zout low.
but now we have a few more options open for us. now all you need is a simple tapped winding. I'll have to think about (lamely attemt to interpolate) what theory says for this situation and get back to you. (wh
The inductance seems to be OK Dave, 'cause I changed the line stage... see below. Could I have a couple more bobbins thusly wound? Should be "finals".
not so fast there :-) I know we have to draw the line somewhere, so given the info at hand, lets use theory and previous experience to finalize the design and i'll do a pair which will essentially be #4 with a tweeked balance winding (maybe 1.5dB each way??) and then you can compare the finished pair to the old standard #4 and hopefully call it case closed.

Made all the difference in the world. With the lower rp of the 5687, the inductance is just fine. In fact, the way the transformers are wound, I may end up putting the halves of the secondary in series if I want more gain.
i would love to see some FR measurements compared to listening results for those two situations. If you do it, please listen first then measure, since we all know how measuremeants can bias us.
Attenuator... the one limitation I found I noted above. With tubes that require relatively low bias voltages
is it low bias, or high gm?? i guess high gm suggests low bias, but it also suggests the other evils.

i have a linestage with current compensated TVC loading a 12B4 driving an amp with a 437A input tube (i know... bonehead design placing the 12B4 prior to the 437) and occasionally one channel pops on attenuation. Maybe i should swap tubes and see if i can track down the cause.
With high bias (5687 at 7.5V qualifies, as would 76, essentially all DHTs etc) tubes, the grid current is low and constant so there is no issue.
do you think the issue would show up in a 5687 biased at -2V ???
Also found the overall loudness effect less subtle on the "final" versions. Ended up changing the interleaving to 2:1 which works just fine.
i think this is an important issue... what does 2:1 mean? I'm guessing given the change in source Z the curve changed a bit. Knowing what caused it and being able to fix and predict it "quickly" will keep the concept alive. Otherwise it just becomes tweeky.
whatchya wanta tackle next? The tone controls?
you tell me... essentially you are doing 90% of the work... and i think we are 90% of the way there with the 3:1's (at some point i should do a pair of amorphous versions just because i would love the ears on experience.

interestingly enough, the amorphous C-core i would use would also fit a slim stack of 80% nickel to attempt to close the circle a bit.

the tone controls will be a bit easier, since we will have to make a decision on design and do it. tweeks at that point of the game will just drive us nuts for something that may not be needed.

dave
dave
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

not so fast there I know we have to draw the line somewhere, so given the info at hand, lets use theory and previous experience to finalize the design and i'll do a pair which will essentially be #4 with a tweeked balance winding (maybe 1.5dB each way??) and then you can compare the finished pair to the old standard #4 and hopefully call it case closed.
Ya got a deal there. The 1dB each way difference is, as I mentioned, subtle, but the balance principle seems to work just fine. I would expect the 1.5dB would be essntially "perfect". OTOH, what doesn't work so well yet (so there's no schematic in here yet) is the mix-to-mono. (the 6th position on the balance switch in my setup).
Quote:
Made all the difference in the world. With the lower rp of the 5687, the inductance is just fine. In fact, the way the transformers are wound, I may end up putting the halves of the secondary in series if I want more gain.

i would love to see some FR measurements compared to listening results for those two situations. If you do it, please listen first then measure, since we all know how measuremeants can bias us.
Agreed. Been more in listening than measuring mode... I have not re-measured things since I switched to 5687... may try altering the secondaries, listening, then doing measurements in both cases.

is it low bias, or high gm?? i guess high gm suggests low bias, but it also suggests the other evils.

i have a linestage with current compensated TVC loading a 12B4 driving an amp with a 437A input tube (i know... bonehead design placing the 12B4 prior to the 437) and occasionally one channel pops on attenuation. Maybe i should swap tubes and see if i can track down the cause.

Quote:
With high bias (5687 at 7.5V qualifies, as would 76, essentially all DHTs etc) tubes, the grid current is low and constant so there is no issue.


do you think the issue would show up in a 5687 biased at -2V ???
From what I've been able to measure, it is a residual grid current issue. For any tube (but magnitudes and inflection point change on a type to type basis and to a lesser extent on a device to device basis for the same type). At zero bias, there is some grid current flow, which decreases to exactly zero at somewhere around -1.5 volts. It changes direction and increases until about -3 volts then gradually decreases to zero at high biases. I expect 5687 biased at -2 volts would show the same TYPE problem but to a lesser degree (lower mu). Hi mu devices get hit with a double whammy... the hi mu, by definition sort of, requires low bias voltages, since cutoff occurs at lower grid voltages, and the hi mu then multiplies whatever change you forced on the grid (resistance change during switching) by mu.
(well, that's true of lo mu, but the same impedance induced change on the grid produces a "pop" output proportional to mu).
Quote:
Also found the overall loudness effect less subtle on the "final" versions. Ended up changing the interleaving to 2:1 which works just fine.


i think this is an important issue... what does 2:1 mean? I'm guessing given the change in source Z the curve changed a bit. Knowing what caused it and being able to fix and predict it "quickly" will keep the concept alive. Otherwise it just becomes tweeky.
2:1. too much shorthand. two E swap direction 2 E swap direction 2 E swap direction......

I had mentioned this before... the final attenuators we ended up with were changes in core as well as windings. The "original" were the thin lam and I had stacked 10 E at a time. With the final set, I started 5 E then swap direction, then moved to 4 Es then 3 Es then finally 2 Es, which I liked the sound of best in the loudness position. It's slightly more subtle than the original. The 3 Es were probably very close to the original shot with the other lams.
you tell me... essentially you are doing 90% of the work... and i think we are 90% of the way there with the 3:1's (at some point i should do a pair of amorphous versions just because i would love the ears on experience.

interestingly enough, the amorphous C-core i would use would also fit a slim stack of 80% nickel to attempt to close the circle a bit.
Cool. Be glad to try that. Lets continue the other 10% of the line stage transformers then tackle the tone.

Don't forget... the original offer still stands. Any time you (or anyone lurking on these threads) wants to have a listen, I expect this thing is transportable successfully. :D

Steve
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

sbench wrote: Ya got a deal there. The 1dB each way difference is, as I mentioned, subtle, but the balance principle seems to work just fine. I would expect the 1.5dB would be essntially "perfect". OTOH, what doesn't work so well yet (so there's no schematic in here yet) is the mix-to-mono. (the 6th position on the balance switch in my setup).
so i'll go for 1.5dB windings for the balance. I'm still not sire how to place them for an improvement. I think going to a lower DCR, might open a can of worms with the positioning of the winding.

the mix to mono might merit a new topic, since i suspect magnetic summing of stereo to mono is very similar to a real world class A PP output stage.

I had mentioned this before... the final attenuators we ended up with were changes in core as well as windings. The "original" were the thin lam and I had stacked 10 E at a time. With the final set, I started 5 E then swap direction, then moved to 4 Es then 3 Es then finally 2 Es, which I liked the sound of best in the loudness position. It's slightly more subtle than the original. The 3 Es were probably very close to the original shot with the other lams.
did you come up with a general feel for any difference between the thick and thin lams??? counter to conventional thinking, several people have voiced a strong preference for the thick lams javascript:emoticon(':shock:')

Cool. Be glad to try that. Lets continue the other 10% of the line stage transformers then tackle the tone.
i think we can start the basics of the tone while we tweek the other.

i'll think a bit about the next rev of the 3:1 and ring you about balance winding placement.

dave
Raj Gupta
Posts: 37
Joined: Fri Feb 11, 2005 2:19 am

Post by Raj Gupta »

I think one way good broadcast equipment sums to mono is a 1:1:x transformer, with l and r each getting one of the 1's and taking the summed output signal from the x winding. The other way is some variation of a mixer setup.

-j
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

so we are at version #7 :-)

i think we have made some moves in the right direction from a measurement perspective, and am beginning to think that we could be nearing the end of this journey for a bit. (for the time being at least) I'm still turning up some quirks as i wind and measure these things particularly wrt the balance winding, but feel i may need a fresh perspective.

#7 is the first to behave in a manner more in line with what i have experienced with my outputs. Namely something that approaches a bessel rolloff prior to the first resonance. Reversing the ground on both windings extends the bandwidth a good amount but results in a much steeper rolloff.

I am somehow more comforted by the more gradual change even if it does begin much earlier.

another interesting note is the positon of the balance winding has a dramatic effect on the performance. In the plots, for the +1.5 the winding is placed in series before the main winding, and for the -1.5 it is also placed at the start, but the overall phase is reversd !?! (i think thats how it was wired, my head really starts to hurt with all of those clip leads floating around). The point is, phase and position of the windings has a drastic effect on the measurements. (note i'm talking about behavior above 100Khz)

another interesting thing to note is simply touching both sides of the output does wonders to damp any oddities. (using your body as a load) I have noticed this behavior before, and have been unable to replicate this type of loading with any RC combo. the results are easily seen when looking at a 10K square wave.

#4 and #7 are quite similar to each other, and my #7 "reversed" looks quite similar to your #4 measurement, so maybe we are chasing out tails here for little gain. (i did note that in your plots above, the #4 plot seems to change from the main graph to the balance winding graph ?!?)

let me know how you want to proceed. I'm guessing another rev 7 would make a matched pair to compare to rev#4. Hopefully i can get that done over the weekend and ship them with the tone control chokes early next week. Then while you play, i can get to finishing up some stuff for the c3 (oops c4) guys.

dave
Attachments
version #7  the primary to secondary phase is kept the same and the ground is reversed for the main traces, and the +- were referenced to the "traditional" hookup.  The wider bandwidth of the "reversed" hookup has a much steeper slope to its rolloff sugge
version #7 the primary to secondary phase is kept the same and the ground is reversed for the main traces, and the +- were referenced to the "traditional" hookup. The wider bandwidth of the "reversed" hookup has a much steeper slope to its rolloff sugge
image007.gif (11.5 KiB) Viewed 26058 times
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

let me know how you want to proceed. I'm guessing another rev 7 would make a matched pair to compare to rev#4. Hopefully i can get that done over the weekend and ship them with the tone control chokes early next week.
Lets go with that. That is, 2 #7s and tone control chokes. At present, it looks like we're doing fine.

By the way, I can turn the "sound" of one of #3 into that of #4 by altering the "electronics" driving the magnetics. That is, by using half a 5687 on one side and paralleled 12BH7 sections on the other, the resultant stereo is actually well balanced.

Steve
Post Reply