801 parafeed choke

Design and use of Chokes for PS, anode and filament use.
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

801 parafeed choke

Post by sheldon »

I made a couple of resistor loaded 801 amps based on the design in Steve Bench's site. Needing more than the 400mW for an application, I converted one to a regular parafeed style amp, that can go into A2 for higher signal levels. The amp is shown attached.

So far, so good. But when I measured the output, I saw some pretty extreme ringing with a 1kHz square wave. Even more disturbing were some extreme distortion (looked like 3rd harmonic) at frequencies between about 15 and 22kHz. I investigated along the circuit, and subbed in and out various parts. It's clear that the plate chokes themselves are the cause of the ringing and harmonic distortion. A more detailed description the amp and output is here: http://www.diyaudio.com/forums/showthre ... genumber=1

So, I have a number of options here: A regular SE transformer, a different plate choke, or maybe a bifilar wound plate choke. That last one appeals to me. What do you think?

To jump the gun a little, the current plate chokes are on EI 112 cores. I happen to have some 112 lams left over from another project. How bout a core wound for those lams? Would make it easy to fit to the chassis without mods. This part of course is nice, though not essential.

Sheldon
Attachments
801A2 parafeed power drive.JPG
801A2 parafeed power drive.JPG (28.36 KiB) Viewed 40275 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

what happens if you replace the 10K OT with a 10K resistor and look at the square wave and the harmonics? This should easily isolate the choke as the culprit. I find it hard to believe that a choke with the specs that you describe is the cause of the behavior you are seeing.

What did you do to "fix" the response / how did you isolate the choke as the problem?

dave
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

Thanks for the reply. Yes, I did that. No change. I tried a completely different OT too. I got basically the same outcome. I wrote the following to someone else I was corresponding with:


"Reversing the choke leads didn't change much. It may have changed the amplitude a little but the basic shape of the 1kHz square wave didn't change and the roll off at 15k or so remained about the same.

I tried 1k and 6.8k in series with the choke. The higher value may have changed the ringing amplitude some, but not dramatically. And the roll off also didn't change much.

A 3.9nf cap in parallel with the choke reduces the ringing by a factor of 3 or more, and rounds the leading edge of the wave. But it doesn't change the frequency or affect the point of the output roll off much.

I tried 1k and 4.7k on the grid. No change. Interesting sidelight is that with the long internal posts in the 801 tube, the resistor can only go so close.

So, it's looking like this choke may not be really suitable for this amp. Right now, it's not a problem because I'm only using it between about 200 and 1500Hz, but eventually I'd like to use it from about 1K up. How about a bifilar wound choke and eliminate the parafeed cap?"

After a little more thought and research, I'm guessing that the bifilar idea might not be workable, but I'm open to any solution. I don't need the bottom end for this amp, so am willing to sacrifice some there for good performance mid and up top.

Here's a pic of the distortion around 18kHz, and the 1kHz sqwave.

Sheldon
Attachments
IMG_0839.jpg
IMG_0839.jpg (53.7 KiB) Viewed 40215 times
IMG_0843.jpg
IMG_0843.jpg (25.25 KiB) Viewed 40215 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Yikes!

You dotted all the T's and crossed all the I's and it seems as if the choke is indeed the culprit. I have to admit this is the first time i have seen such behavior and am scratching my head for a possible cause.

Going beyond that... Initially i was going to reply that the bifilar probably wasn't the best approach for a situation where power was involved but now given that you don't need much below 200hz and the top end is what matters i may reconsider. Right now with 125hy's you have bass down low and it is obvious you are paying dearly with it on the top end.

The nice thing about the bifilar is you can start with it as a choke and then move to the IT approach so it serves as a good learning experience. I just did a pair of units for another member and lightly loaded, they had nearly the same frequency response wired as an IT as they did as a choke. Your near 500V B+ is a slight issue wrt suitable wires i have at hand but i'm sure we can figure something out. I'm one of those sick puppies who giggles when you put 4KV across a bifilar trannie and the only result is a spark across the end of leads that are 3/16 an inch apart

dave
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

Thanks Dave,

I'm open to any and all approaches - regular SE, parafeed, exotica like bifilar chokes.

When you mention moving to IT, do you mean for this amp, as a replacement for the fet follower?

While you are pondering solutions to the choke issue, here's another transformer to think about, though this one is not a problem. I left one of the stereo amps in its original resistor loaded form. The OPT's are 12k versions from Electra-print. They look like they would be good for 5W or so. I think that they are just adapted from a pp design, complete with a center tap. Actually, no complaints with the sound or measurements but this amp could be a good candidate for exotica. One doesn't see many purpose built 400mW OPT's at 12k. I may use this amp for the tweeter, or it's quiet enough for headphone duty. Whaddya think?

Sheldon
Attachments
801ampfinal.GIF
801ampfinal.GIF (10.55 KiB) Viewed 40126 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

hey,

by moving to IT's i was referring to wiring the bifilars aseither plate chokes or IT's and listening both ways.

indeed a 1/2 watt 12K OT for HF use only opens up a lot of doors typically held shut by the "you must design for full power at 20hz camp"

dave
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

dave slagle wrote:by moving to IT's i was referring to wiring the bifilars aseither plate chokes or IT's and listening both ways.
If you think the bifilar choke would work, I'd give it a shot. What size core would you use?

Using as an IT is interesting, but would require a much beefier input stage no?
dave slagle wrote:indeed a 1/2 watt 12K OT for HF use only opens up a lot of doors typically held shut by the "you must design for full power at 20hz camp"
That would be fun to try too.

Offline for arrangements?

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

Post by dave slagle »

If you think the bifilar choke would work, I'd give it a shot. What size core would you use?

Using as an IT is interesting, but would require a much beefier input stage no?
for this case (bifilar used as OT to feed the 12K) the DCR needs to be kept down since it will be working into a 12K load. In order to avoid ending up with something that weights 12 pounds, i'd probably end up using something around an EI-100 in amorphous.

I'm not sure i follow the need for a beefier drive, from the 801's POV the IT and the plate choke should appear the same.

WRT the 12K... if you don't need the power, you could consider trying something like 24K. The 801 is similar in Rp to the 75TL and when the guys in colorado played wiht loads they ended up trying 5,10,20, 40 and 80K and their preference was for 40K. They were series feed from 200hz up, but your situation (at this point) allows for some interesting games ot be played full range with either parafeed or the 1:1 feeding the OT.

dave
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

Dave, I consider myself reasonably intelligent, but in these matters mostly ignorant. Dangerous combination for all concerned, but I think I'm starting to get it.

Just to make sure we're talking about the same things here: The choke/IT is for the 5W A2 amp. It finally sank in that you weren't talking about an interstage transformer between the 6SN7 and 801, but switching between its use as a plate choke or as an interstage transformer between the ouput tube and output transformer. I hadn't been exposed to the concept of that being an IT. OK, sounds fun. FWIW the OPT I'm using for this amp is a nickel core EXO-100 from Magnequest.

Switching gears, the half watter is for the resistor loaded amp, ala Steve Bench http://members.aol.com/sbench/inp_dri.html. Lots of choices here. I could go series feed for a conventional SE amp. That has some attraction, as I haven't done one of those yet. It would be easy to modify the power supply to reduce the B+ by changing the input cap and increase the bias current to get the same 0.5W range with a higher winding ratio. Or, keep the resistor load and go with 24k or something. If I did that, I might want to stay full range so I could use it for headphones or a small office system. Getting a little low on juice for the main system with the high winding ratio. Possible to do a little compromise by setting it up for series feed, but leave the option to take out the gap and try it cap fed from the resistor too? That would give me a few things to play with. If that's feasible, I'd give it a go.

Switching gears again, there is another way I'm gonna try using the low watt amp if I need more juice. I've already built the FET amp shown in this concoction: http://www.diyaudio.com/forums/attachme ... 1182731770.

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

Post by dave slagle »

SO for the 5W A2 you can try a plate choke / IT combo.

The 1/2 watter can be done so a single design could be used SE or parafeed, but going limited/full bandwidth would require a different coil design.

the final Schematic is interesting. Are the outputs operated class A? What is the impedance of the transformer?

dave
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

Ok, let's take a shot at the Choke/IT for the 5W amp to start with, and I'll ponder the 1/2 watter a bit.

Regarding the FET follower amp, yes by my calculations I'm operating the outputs in class A. I'm not sure the impedance of the OPT, but here's the description: It's trifilar wound with 0.8mm wire on a square stack EI-112 bobbin. Total turns each wire is about 150. Laminations are M6. Two windings in series for the primary, and the third for the secondary. A more complete description of the amp is in the last few pages of this thread: http://www.diyaudio.com/forums/showthre ... genumber=1

Sheldon
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

Oh, I forgot to add that the trifilar tranny was gapped with a 0.075mm (0.003") shim. I wasn't too concerned about low frequency performance and wanted to account for any imbalance.

Sheldon
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

dave slagle wrote:the final Schematic is interesting.
So while you are winding those chokes how bout pondering this one. As it is now, I have a LL1674 as the input transformer. It's configured with the primaries in parallel, so it steps up 1:4 as a phase splitter. I can drive this directly with a low impedance (75R) preamp, and it works well.

I was thinking of driving it with the 801 amp. But thinking some more, I'd like to do it all in the same chassis. So, maybe a 3A5 (little dual DHT - not stuck on that tube, but something like that would be cool). I was originally thinking I could turn the Lundahl transformer around, but testing with a signal generator, is not promising. I could do something like the attached, which would be straightforward. Maybe the center tap is not necessary and one could substitute a virtual CT to make the transformer easier?

Or maybe get wild and use a the input dual triode as a differential amplifier with a primary on each plate? Wonder if I could use AC on the filaments then.
Attachments
Zeus Cascode 3A5 driver.GIF
Zeus Cascode 3A5 driver.GIF (4.56 KiB) Viewed 39832 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

try # 1 is done...

the L vs. I plots are below. The inductance is a bit lower than i had hoped, how much current are you running?

Driven as an IT with a 4K source and a 12K load we are -1dB at 70K and -3dB at 137K with a few slight 0.2dB wiggles between 20K and 70K. I'd expect the plate choke behavior to be the same.

dave
Attachments
Picture 7.png
Picture 7.png (35.24 KiB) Viewed 39817 times
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

Hi Dave,

I'm running just under 30mA. But, I favor a separate amp for the low end anyway.

Sheldon
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

sheldon wrote:
dave slagle wrote:the final Schematic is interesting.
I could do something like the attached, which would be straightforward. Maybe the center tap is not necessary and one could substitute a virtual CT to make the transformer easier?
Hmm. Playing with a toroid got me thinking about this one. I guess that with the high Zout, a regular SE to PP might be a bit tough. Could make it easier this way?
Attachments
Zeus Cascode 3A5 fet follower driver.GIF
Zeus Cascode 3A5 fet follower driver.GIF (5.26 KiB) Viewed 42111 times
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

dave slagle wrote:try # 1 is done...

the L vs. I plots are below. The inductance is a bit lower than i had hoped, how much current are you running?

Driven as an IT with a 4K source and a 12K load we are -1dB at 70K and -3dB at 137K with a few slight 0.2dB wiggles between 20K and 70K. I'd expect the plate choke behavior to be the same.

dave
I just picked these up at the post office, opened the box and Whoa. If I can paraphrase a line from one of the Crocodile Dundee movies; "That's not a choke, THIS is a choke".

Technical question: Using it as a 1:1 transformer seems straight forward. As a choke, windings parallel, right?

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

Post by dave slagle »

yup.. that is the hookup method.

dave
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

OK, I had already assumed so and did some quick testing with the scope. Both ways look just fine. None of the upper frequency distortion seen with the other choke. Bandwidth is good. Both methods start to show distortion at full power (about 4.5W) around 100 Hz. That may reflect the limits of the OPT, as I think Mike at MQ rates this one at about 3W. Fine in any event, as I biamp (at least) for higher power configurations. I get about 0.5dB down at 20kHz with the choke, about 0.35 down when transformer coupled.

I fabricated a couple of simple brackets to fit the mounting holes on my chassis. Next day or two, might get a chance to do some listening.

Thanks Dave,
Sheldon

P.S. Hope you do this for fun, cause looking at these things, I can't imagine doing it for the money.
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

Ok, here's the latest configuration for listening tests. I changed the load resistor on the FET follower to a CCS, otherwise only the transformer and output configuration change. One curious issue is that I get oscillations if the input is left with only a 300k resistor to ground. Didn't do that before. I suspect the CCS combo is more sensitive to that. No problem with a low Z input, so I'll sort that later.
Attachments
801 A2 Bifilar T coupled.JPG
801 A2 Bifilar T coupled.JPG (29.89 KiB) Viewed 42070 times
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

A little additional testing shows the frequency upper limit is determined by the OPT - the 3dB point is around 50kHz with parafeed, about 35k as a transformer. Some difference between channels, which appears mostly to be differences in the two OPT's. I need a new sound card to play around and look at distortion, but it sounds quite nice.

Here is this iron rich little amp (well endowed with copper too):
Attachments
801slagle.jpg
801slagle.jpg (53.45 KiB) Viewed 42003 times
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

front
Attachments
801slaglefront.jpg
801slaglefront.jpg (51.4 KiB) Viewed 42001 times
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

So far, I'm liking the bifilar plate transformer version. A little better bandwidth on the top end, essentially no distortion above 4th order (the parafeed version has a trace of 5th - down over 100dB), fewer parts. Can't say I really hear a dramatic difference in sound, but I haven't done back and forth AB testing, but it sounds good.

Sheldon
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

sheldon wrote:Ok, here's the latest configuration for listening tests. I changed the load resistor on the FET follower to a CCS, otherwise only the transformer and output configuration change. One curious issue is that I get oscillations if the input is left with only a 300k resistor to ground. Didn't do that before.
Bad solder joint.

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

Post by dave slagle »

Have you done a frequency sweep of the bifilar version at full output? I'm just curious if we kept the copper losses low enough to avoid the level shift as you transfer from magnetic to capacitive coupling.

Also, earlier you said the parafeed was -3dB @ 50K ad the bifilar -3dB @ 35K but above you mention the bifilar had better bandwidth.

My testing showed me the iron measured essentially the same both ways and would be curious as to the cause of the measured difference since at the frequencies in question the parafeed cap is essentially a piece of wire anyways.

dave
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

dave slagle wrote:Have you done a frequency sweep of the bifilar version at full output? I'm just curious if we kept the copper losses low enough to avoid the level shift as you transfer from magnetic to capacitive coupling.

Also, earlier you said the parafeed was -3dB @ 50K ad the bifilar -3dB @ 35K but above you mention the bifilar had better bandwidth.

My testing showed me the iron measured essentially the same both ways and would be curious as to the cause of the measured difference since at the frequencies in question the parafeed cap is essentially a piece of wire anyways.

dave
Ok, you caught be being sloppy with the measurement and analysis. The first set of measurements with a generator and scope. I adjusted the input signal at 1k to just under visible clipping and then manually swept the frequency range until I got a drop of 3dB on the upper end. I compared on channel bifilar with the other channel in parafeed. I didn't account for the fact that the chokes or OPT's could vary a bit. Later, IIRC, I checked both chokes with the same OPT and didn't see a difference so the difference I first reported between bifilar and parafeed could well just be OPT variation.

In the recent case, I used a soundcard at lower volume (maybe 1/4 full output voltage) and saw a slight difference between the bifilar and parafeed, in favor of the bifilar. But that is a pretty low resolution plot and could be a mirage. So, without more careful measurement, I would have to say that we are probably within experimental error.

I did not see any evidence of a level shift when I swept with the scope or soundcard.

Sheldon
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

from the top:

bifilar THD

parafeed THD

bifilar frequency plot

parafeed freq

Green trace is soundcard control. Plots are a bit noisier than I see with the amp all buttoned up.

Edit: I apologize for not scaling these to expand the frequency resolution. Note, that if you look close, there is a bit of rising response in the transformer coupled output, starting at about 5kHz. Maybe that also creates an illusion to make it look like the roll off point is a bit higher, when it actually isn't. I see the same result with the other channel as compared to the parafeed set up. Close enough for music, I'd say.

Sheldon
Attachments
A2 left freq.png
A2 left freq.png (5.49 KiB) Viewed 41908 times
A2 trans left freq.png
A2 trans left freq.png (5.43 KiB) Viewed 41908 times
A2 left THD.png
A2 left THD.png (9.99 KiB) Viewed 41908 times
A2 trans left TH.png
A2 trans left TH.png (10.21 KiB) Viewed 41908 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

thanks!

that slight rising response looks to be the transition i would expect and it seems we kept the DCR's down enough. You mentioned this was at about 1/4 power and if this is indeed the transition as you go to full power the HF "bump" should become more pronounced.

how do you like the software you are using?

dave
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

I did a little more testing. With the scope and signal generator (and DMM), I don't see significantly more bump as I scan from 1k up to 20kHz, or down to 100, at maximum power.

I put a pot on the input to the sound card, so I could bump up the amp output as much as possible without clipping in the sound card. It's an M-Audio transit (can be had for $50 on E-Bay, sold as the Groove Lab). The maximum specified output of the card is 0dB, so I assume 1VRMS. My amp clips at 2.5V in (6V out), so measurements taken down 8dB. The previous measurements were down a further 4.5dB, so upon review more like 1/16 power. Anyway, here's the expanded graph taken at full sound card output.

If I want to go higher, I need to make up a little preamp.

The RMAA freeware is easy to use for these measurements, but it's not real flexible. I guess ARTA is well thought of. When I get the chance I'll play with it. Problem is, I don't like to spend a lot of time learning new software, so I procastinate.

Sheldon
Attachments
left trans max card output.png
left trans max card output.png (6.58 KiB) Viewed 41877 times
sheldon
Posts: 24
Joined: Sun Sep 02, 2007 4:54 am
Location: San Diego

Post by sheldon »

I cleaned up my test system. Here's THD at just under 1W output. It looks like this until about 3W, then the higher level harmonics start to increase. Interesting is that when I tried a cascode DN2540 current source on the 6SN7, the distortion increased slightly. This was true for both at 1W and 3W. I had expected the distortion to be a little less, especially for the larger swing.

Seems that Steve Bench guy knows what he's doing, as this is adapted directly from his design.

Sheldon

edit 1/21
Attachments
A2 trans left freq.png
A2 trans left freq.png (9.9 KiB) Viewed 41764 times
Post Reply