initial gap selection

that colorado rocky mountain high.
Post Reply
dave slagle
Posts: 2102
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

initial gap selection

Post by dave slagle »

whew,

This is the third time typing this in so i'll keep it short. there was a minor posting bug which I hopefully have fixed.

I need to set the initial gaps for the outputs and the 2:1's for shipping so attached below are the L vs I plots for the appropriate range.

I'll get this post up with the plots and add more below.

my gut feel is the two stages will be about -3dB at 40hz and -1dB at 80hz. the selection of the bypas cap on the stages will allow you to tweek this a bit.

more to follow.
Attachments
This is the Low turn 2:1.  It is well behaved out to 100K and for your use i'd suspect a .006 gap will net yo a nice 25ma operaing point.  the 25hy with the Rp of the RS241 should put you -1dB at 50-60hz.
This is the Low turn 2:1. It is well behaved out to 100K and for your use i'd suspect a .006 gap will net yo a nice 25ma operaing point. the 25hy with the Rp of the RS241 should put you -1dB at 50-60hz.
image001.gif (8.61 KiB) Viewed 14076 times
for this i'd guess the .015 gap would be a good choice.  The 30hy will give you a slightly lower -1dB point than the driver but since this is a power stage the extra inductance should help.
for this i'd guess the .015 gap would be a good choice. The 30hy will give you a slightly lower -1dB point than the driver but since this is a power stage the extra inductance should help.
image001.gif (9.81 KiB) Viewed 14076 times
Johnny
Posts: 62
Joined: Wed Jan 26, 2005 3:43 pm

Gaps

Post by Johnny »

Hi Dave,

Looks good. Do you have an initial guess where the chokes should be gapped? Can't wait to try this stuff out!


Johnny
Steve Kaufman
Posts: 35
Joined: Wed Jan 26, 2005 6:00 am
Location: Fort Collins, Colorado

Post by Steve Kaufman »

Hi Dave

We're getting pumped about answering all these new questions. James is laying out the four chassis. Before he can order materials he needs to know the dimensions of the footprints of the output and IS trannies as well as the chokes. Would you mind giving me a rough measurement of these when you get your hands on them. This will give James some extra time to get things started.

Thanks

Steve
Steve Kaufman
Posts: 35
Joined: Wed Jan 26, 2005 6:00 am
Location: Fort Collins, Colorado

Post by Steve Kaufman »

Dave

There is one more thing I neglected to clarify with you. The "temporary" cores you are sending we figured to be amorphous as the final ones will be. If this is not the case it will create an additional variable. Will they, in fact, be amorphous?

:-O

Steve
dave slagle
Posts: 2102
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Steve Kaufman wrote:Hi Dave

Before he can order materials he needs to know the dimensions of the footprints of the output and IS trannies as well as the chokes.
the outputs are 5.5 inches cubed, and the IT's are 3.25 inches cubed. The chokes are essentially the same size as the IT's (one of the dimensions is a bit smaller due to the lack of pins.

There is one more thing I neglected to clarify with you. The "temporary" cores you are sending we figured to be amorphous as the final ones will be.
yes they will be the same cores. I did some calculations and moving from the temporary cores to the final cores will give you a 2cm shorter magnetic path which for the gaps we are talking about (.006 in) translates to an increase of inductance of about 1-2hy's when the final cores are put into place so in my opinion the "variable" will be non-existant.

dave
Steve Kaufman
Posts: 35
Joined: Wed Jan 26, 2005 6:00 am
Location: Fort Collins, Colorado

Post by Steve Kaufman »

Hi Dave

The iron arrived in two parcels today. John and I will get down this weekend. Stay tuned!

Thanks

Steve
dave slagle
Posts: 2102
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

gaps for the plate chokes.

Post by dave slagle »

hey guys,

glad the stuff made it there in one piece :-)

here are the L vs. I graphs for the new choke bobbins to try as a parafeed drive for the 75TL.

the easiest way to identify is by DCR, but it is a ballpark number... so for the record here are the corresponding wire sizes.

~700=#33
~425=#32
~260=#31
~175=#30

i'm embarrassed to admit i had the plots for all three, but somehow accidentally wrote over the #32 version :-(

the last plots are form the original tests before you guys became gap crazy and it represents the only data i have for the 700R bobbin you have on hand.
Attachments
the gap for both bobbins is .006
the gap for both bobbins is .006
image001.gif (5.15 KiB) Viewed 14031 times
image001.gif
image001.gif (15.24 KiB) Viewed 14032 times
image001.gif
image001.gif (9.78 KiB) Viewed 14032 times
Steve Kaufman
Posts: 35
Joined: Wed Jan 26, 2005 6:00 am
Location: Fort Collins, Colorado

Transformer/choke trials

Post by Steve Kaufman »

Hi Dave

With the various transformers and chokes you sent us recently we attempted to optimize the sound we can obtain by altering the number of windings in the cores and the gap betweent he C-cores. In each case we strive to reach a sweet spot in which the highs are most extended, the midrange and highs are most detailed and open, and the bass response is not compromised. In general, widening the gap and/or reducing the number of windings (lowering the DCR) tends to improve the mids and highs and, if taken too far, reduces the bass extension, "thinning out" the bottom end. The opposite is the case with reducing the gap and increasing the number of windings. These assumptions we have made are based on affecting the inductance. The goal is to maintain a constant inductance while reducing the DCR.

These parameters were tested in John Pessetto's 75TL single ended stereo amp with RS 241s in the driver stage. The gaps were set using Artus shim stock. The first run employed the "reduced turn"(approx 20%) amorphous core output transformers with the gap set at 0.015" (your setting out of the box). The driver stage was configured in parafeed with a Magnequest B7 parafeed choke, a 4uF GE "oiler" cap, and your "full turn" amorphous core choke gapped at 0.06". The sweet spot was achieved with a gap of 0.045". (range 0.015-0.060").

The next change was to "full turns" bobbins in the output transformers. The sweet spot was now reached with a gap of 0.025" (0.020-0.060).

Comparing the output trannies with " full turn" vs. "reduced turn" bobbins optimized with respect to gap, the "full turn" configuration sounded faster and tighter with more energy without compromising the bass response or high end extension and air.

The next change was to optimize the driver stage choke using the full turn (7000) bobbbins. We started out with the 0.025" gap you set. The sweet spot came at a gap of 0.030'(0.020-0.060).

We then planned to to substitute different turns bobbins in the driver stage choke but found that the bobbins we received were too large for the frame. We later figured they could have been used in the frame for the interstage trannies we were about to try but at that point we switched the parafeed arrangement to a 2:1 interstage trannie, eliminating 2 chokes and a capacitor.

The interstage trannie came with reduced turns bobbins and a gap set to 0.0025". Relative to the parafeed arrangement this set-up was far superior. It was as if a thick veil had been lifted. Detail improved. The sound was much more natural with increased air and tighter bass. We closed the gap down to 0.002" with slight improvement in the bass but no compromise of the mids and highs. We stopped there because we were into "tissue paper" gaps here.

The next change involved changing to the full turns bobbins for the 2:1 interstage trannie. We set the initial gap at 0.002" and the sound was shifted toward the bass, with some loss of high end extension and air. Changing the gap to 0.004" restored the balance and brought back some of the magic of the low turns bobbins. We dropped the gap to 0.003" resulting in the most balanced sound. As in the case of the output trannies, the "full turns" bobbins presented a faster, more dynamic sound than the "reduced turns" version.

We retried the reduced turns bobbins in the output trannies before finishing and found the result to be consistent even with all the other changes made to this point.

We rechecked the chokes in my preamp for gap. Since the "full turns" version of everything seemed to work better we reoptimized the gap and found no change from what we had arrived at months ago with a different power amplifier. We ran out of time and did not try the reduced turns bobbins. We were also more concerned about losing the low end in the preamp which has to reproduce the full frequency range, unlike the power amplifier which only has to reproduce down to 100 Hz.

The end result was beautiful sounding. After talking with you earlier, the "full turns" bobbins John brought over were optimized for his 16 ohm speaker load rather than my 8 ohm load. At some point, after my amps are built, it might be worthwhile trying a "full turns" version of those bobbins optimized to 8 ohms. Also, it might be fun to try a set of "nanocrystalline" cores in the output transformers.

All in all this was a pretty interesting experience. It would be great if this information would be helpful for you to better predict the best winding pattern and gap selection for the magnetics you supply. Most DIYers wouldn't think of doing this fine tuning themselves. It makes a huge difference.

Thanks

Steve
dave slagle
Posts: 2102
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Hey guys, Thanks for the report!
Comparing the output trannies with " full turn" vs. "reduced turn" bobbins optimized with respect to gap, the "full turn" configuration sounded faster and tighter with more energy without compromising the bass response or high end extension and air.
The whole "full turn" and "reduced turn" may get a bit confusing so I'll give the history and some observations. The first pair of outputs I sent you guys to listen ot were designed for a 300B so an 8 ohm load would reflect back 5K. Even though the Rp of the 75Tl is 5K I felt we would be OK since johns speakers are 16 ohms and he crosses over below 150hz or so. This gives a 10K reflected load and still left plenty of inductance to deal with the 5K source above say 100hz. The large gap initially chosen by john for the "full turn" versions prompted us to wonder if reducing the turns could net a further improvement so I wound a version with 1/3 less turns and indeed the results were positive (again remember this is for >100hz operation)

Armed with these two experiences, we needed to come up with a version for SLK's system and I came to the conclusion that since his speakers are 8 ohms it would be wise to increase the turns ratio to 35:1 to keep the same 10K reflected load (2X Rp) and only reduce the turns slightly over what was deemed the "full turn bobbin" SLK still uses active subs but he crosses them over about an octave below johnny so I split the difference between the two.

The fact that the original full turns bobbin sounded best really surprised me. Mainly because in that configuration the Reflected load is = the Rp of the tube it is loading which definitely would raise some ire from many people out there. I will also note that the top end frequency response is loosely based on turns so from best to worst we have:

Johhny's 1/3 reduced turn 25:1
The "failed" 1/5 reduced turns 35:1
The full turn 25:1

So the preferred transformer actually has poorer top end performance when looked at from a simple frequency sweep. So where does this leave us? I come up with two possibilities and they are both quite simple.

Either the extra turns allowed for a better linearity vs. inductance combination in spite of the Rsource=Rload.

Or The Rsource=Rload actually works in this situation. This wouldn't surprise me since the 75TL has tons of spare current and with steve's horns (AG trios) I suspect the amp is behaving in the voltage realm anyways.

At some point it would be interesting to try johhny's 1/3 reduced turns bobbins which and to try a 35:1 full turn bobbin so wr can close up a few of the loopholes :-)
we switched the parafeed arrangement to a 2:1 interstage trannie, eliminating 2 chokes and a capacitor.
Also eliminating a few hundred dollars in parts and a few dozen other possible listening swaps :-)
The interstage trannie came with reduced turns bobbins and a gap set to 0.0025".
For looking at the L vs. I graphs remember you need to double that number to .005 since the gap enters the magnetic path twice.
Relative to the parafeed arrangement this set-up was far superior.
:-) :-) javascript:emoticon(':twisted:')
It was as if a thick veil had been lifted. Detail improved. The sound was much more natural with increased air and tighter bass. We closed the gap down to 0.002" with slight improvement in the bass but no compromise of the mids and highs. We stopped there because we were into "tissue paper" gaps here.
I assume when you say a .002 gap you were referring to a red shim which translates to a .004 gap. An actual .002 gap requires the amber (.001) spacer. I'm just trying to be clear. I have the L vs I plots for those smaller gaps plotted, but I didn't include them in the graphs above since they didn't seem appropriate.
The next change involved changing to the full turns bobbins for the 2:1 interstage trannie. We set the initial gap at 0.002" and the sound was shifted toward the bass, with some loss of high end extension and air. Changing the gap to 0.004" restored the balance and brought back some of the magic of the low turns bobbins. We dropped the gap to 0.003" resulting in the most balanced sound. As in the case of the output trannies, the "full turns" bobbins presented a faster, more dynamic sound than the "reduced turns" version.
See the L vs. I plot for the high turn bobbins below and the attached comments. I think we may be finding a pattern to your gap selection or at least a good way to pick a starting point.
it might be fun to try a set of "nanocrystalline" cores in the output transformers.
Well since the quote for the nano cores for the 2:1's came in at $600, I would expect a set for the outputs to be around $3600 :-)

The commercial aspect of the nano cut cores is still very new and it seems that nano is becoming the new catch phrase. See the graph below showing the nano family compared to the amorphous and steel family. If you want to impress your friends, start talking cobalt nano "hitperm" :-)

Most DIYers wouldn't think of doing this fine tuning themselves. It makes a huge difference.
I think you guys have helped a lot in demystifying the whole process. For some reason people feel they are not qualified to dive into the black box known as a transformer. I fail to see what makes me as the designer more qualified aside from the fact that they can lock the box and throw away the keys. Somehow I fell you guys will make greater strides with all of the keys in hand and simply considering me a locksmith :-)


dave
Attachments
this is the full family oc curves that I plotted for the full turn bobbin.  It is interesting to note that the lines for any two gaps will cross at some point. If you interpolate where the .006 and .004 gap lines would cross it would be at about 25ma and
this is the full family oc curves that I plotted for the full turn bobbin. It is interesting to note that the lines for any two gaps will cross at some point. If you interpolate where the .006 and .004 gap lines would cross it would be at about 25ma and
image001.gif (10.91 KiB) Viewed 14007 times
this represents the family of amorphous, nano, and plain old steel.  The cores from hitachi (metglas) metals are finemet and represent the only version of cut cores that i have found.
this represents the family of amorphous, nano, and plain old steel. The cores from hitachi (metglas) metals are finemet and represent the only version of cut cores that i have found.
nan0.gif (31.71 KiB) Viewed 14008 times
dave slagle
Posts: 2102
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

now that you have listened, below are the freq plots of the low turn vs. hi-turn bobbins.

dave
Attachments
#2 is the low turns and #3 the high turns.
#2 is the low turns and #3 the high turns.
Picture-1.gif (14.69 KiB) Viewed 14005 times
Steve Kaufman
Posts: 35
Joined: Wed Jan 26, 2005 6:00 am
Location: Fort Collins, Colorado

Post by Steve Kaufman »

Hi Dave

The gaps described were doubled, adding the two sides together. Interesting discussion. Didn't realize the nano cores at $600 were for the chokes! Yikes!

Steve
Post Reply