Afterglow - now I really mean it!

pboser · 6222

0 Members and 1 Guest are viewing this topic.

Offline pboser

  • Full Member
  • ***
    • Posts: 118
on: May 14, 2020, 06:22:31 PM
Last year I was all talk about updating my Afterglows.  https://forum.bottlehead.com/index.php?topic=11884.0  But with all this at-home time on my hands I really mean it now.  I dug them out and recalled that one was blowing the fuse. I checked it out and it seems that's still the case. 

In my last thread I was incorrect about which power transformers I have - when one of my original S.E.X. amps lost a transformer I got a pair of PGP8.1 and did the conversion to the Afterglow.  (About 2000).

So where do I start with troubleshooting?  Lucky for me I have one good amp to compare to. And I found my Afterglow Upgrade manual too!

Thanks in advance.  And stay healthy, everyone.
Pete

Peter Boser


Offline Paul Birkeland

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 19319
Reply #1 on: May 14, 2020, 06:24:24 PM
Hello Pete,

Post some photos of what you have and I would be happy to take a look.

Since this kit is so ancient, we will have to talk about things like plates, grids, cathodes, etc., so it would probably be helpful for you to print the pinouts of the 45 and 5965. 

-PB

Paul "PB" Birkeland

Bottlehead Grunt & The Repro Man


Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5751
Reply #2 on: May 14, 2020, 08:02:03 PM
As it happens, one of my quarantine projects lately has been to dig into the history of those days - so far with a focus on the SEX/Afterglow/Paraglow/B-glow/Parabee/Paramour - all of which were built on the same chassis plate(!)

The PGP8 produces more voltage than the original power transformer. The good thing is, you can get more driver headroom, which was pretty marginal; the bad is that it pushed the PSU voltage to where the 450-v caps had a shorter than usual lifetime, and it increased the 2A3 current which already exceeded spec. (That may even be why it's blowing fuses - look for bulging caps. PB has good solutions for that. ) You will need to change out the big 2500 ohm resistor - the actual value depends on whether you keep the original driver or convert to the SRC4S Paramount style.

TFA-204 would be down 3dB at about 40Hz, into a resistive load.

Last year you mentioned a 45 option - but really, the 45 wants to see a 4K-5K load, not 3K. Stick with the 2A3.

If you can fire up the working amp and **carefully** measure the DC voltages at the 2A3 socket pins, that will help along with the pictures that PB mentioned.

Paul Joppa


Offline Doc B.

  • Administrator
  • Hero Member
  • *****
    • Posts: 9540
    • Bottlehead
Reply #3 on: May 15, 2020, 05:24:03 AM
Re the power supply filter, I got one or two Paraglows in for service a few years ago and to upgrade the original 100uF 450V caps I stacked two 220 or 250uF caps with a 350V rating, bypassed with 1M resistors to allow for equalization of the potential across them. Pretty sure that is what PB does now on any Paraglows that come in for service, and probably a lot neater installation than the way I did it. The original caps are absolutely due for replacement after 20 years running near their design limit, and this is a solid long term solution.
« Last Edit: May 15, 2020, 05:30:15 AM by Doc B. »

Dan "Doc B." Schmalle
President For Life
Bottlehead Corp.


Offline pboser

  • Full Member
  • ***
    • Posts: 118
Reply #4 on: May 15, 2020, 03:49:03 PM
Hi gentlemen,

I appreciate you jumping in so quickly.  I was slowed by working from home today, but got to spend some time with the amps later today and came to the conclusion that I might be a dope.

They are both working this evening.  My best guess is that when I was blowing the fuse I was using a stash of 1A (fast blow) fuses, until I ran out of them, and then I borrowed the fuse from the working amp, and I suspect the problem switched amps. Since I ran out of fuses last night I went to my local hardware store, and got their only 1A fuses. I installed one and both amps work.  I suspect a failure of troubleshooting logic. As I said, I might just be a dope.

So, since I have you, and not wanting to feel like I've wasted your time, where do I go from here? (assuming they continue to operate). I have photos of the underside of one amp if that will help.  A number of years ago I switched to a 7K cathode resistor and used 45 (or 46 with an adaptor) output tubes, but I know it's not a real 45 amp, and am planning on switching it back for the 2A3.

I see Doc's suggestion about updating and upgrading the power supply filter caps.  Are there any other parts that need to be updated?
And, as was discussed last fall, I'd be interested in the shunt regulated driver and soft start, if they are suggested.  But I'd like your advice about whether it's worth pursuing. 

Thanks again for your willingness to help. Further proof that this is the greatest forum out there!

Be well,
Pete




Peter Boser


Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5751
Reply #5 on: May 15, 2020, 06:22:55 PM
This interests me because I've been digging into the history, looking at some designs from today's perspective. Looking at this circuit, there are some changes I'd enjoy working on.

* The cathode resistor should be changed. As PB pointed out last year, it should have a much higher dissipation rating, for the sake of reliability. But also, the current through the 2A3 is too high, both for the 2A3 and for the TFA-204. And the PGP8.1 just exacerbates these issues. The socket voltages will give me a way of addressing the right value, even if they are taken with the 7K resistor and the 45 tube.

* Early in the design process, John "Buddha" Camille wrote a critique of the circuit from which (I believe) the design was derived. He made some good points, one of which was that a high-mu driver needed more plate voltage. I came across a copy of that critique just two days ago! Getting the output tube cathode resistor right will help (allowing the driver more voltage to work with).

* A lower-gain driver than the 5965 (mu 47) such as the 5670 (mu=35) or the 6CG7/6SN7 (mu 20) would also help meet Buddha's criterion, and would be very easy to do. They are also even more linear than the 5965 which is already pretty good.

The above changes will leave you with a good historical example of the Afterglow, with (I believe) audibly improved performance and reliability. The following two would be more fun/work, probably sound better-but-different, cost much more money, and diverge further from the original design:

* The shunt-reg Paramount driver would sound better at low power, but the original might overload more gracefully. So it's a choice to think about. You can't just sub it in, the 5670 won't adjust to a low enough plate voltage - unless you change to a 6CG7 as mentioned above.

* At one time mention was made of converting to Paraglow. That's a much bigger project, and would leave you with a homeless TFA-204. But it's possible - you might even squeeze in the MonAmour iron complement.

Paul Joppa


Offline pboser

  • Full Member
  • ***
    • Posts: 118
Reply #6 on: May 16, 2020, 06:26:32 AM
Hi PJ,

It's serendipitous for me that you're interested in the SEX chassis plate history - I'm happy to take advantage of help your interest!

First, I think I'm more interested in making these an appropriately operating historical example, rather than, e.g. converting to a Paraglow.  These aren't my only amps, they are making me more confident (after 20 years) to try my hand at scratch building some other amps, and I usually use them with open baffle speakers, with or without separately amplified bass support, so the bass limitation of the TFA204s is okay with me.

I have Buddha's VALVE articles and am re-familiarizing myself with them but I'm reaching the same conclusion that I did in the past - that they are daunting in their scope, and too complicated for me to try to implement in the confined space of the small chassis.  I am interested in improvement of my Afterglows, so if there are some of his suggestions you think I should add I'll be happy to take your guidance.  But at my skill level it might be better for me to try to implement his ideas in my next-step breadboard or spacious chassis experiments.

I took some measurements of the output tube sockets this morning - with the 7K resistor and a 46 tube in an adaptor.  (If you think I should change one or both of these and re-measure, let me know.)
Pin 1 - 213/235
Pin 2 - 466/462
Pin 3 - 188/220
Pin 4 - 213/235

I referred to the Afterglow conversion manual and see that these are higher than what was expected then. 

So what's next?

Thanks again, and in advance.

Pete


Peter Boser


Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5751
Reply #7 on: May 16, 2020, 09:57:06 AM
Next is, I search my notes and measure some resistances, then I calculate some changes. Should take me a few hours before I can make a concrete proposal. Meanwhile, would you measure your powerline voltage? It may be a couple volts high.

Paul Joppa


Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5751
Reply #8 on: May 16, 2020, 10:53:36 AM
Got some preliminary numbers for the power stage, but some oddities on the driver.

If the amp is stock, there will be a 182-ohm  cathode resistor on the paralleled driver triodes - pins 3 and 8 of the 9-pin socket. That should give 1.56v drop. Do you in fact have 182 ohms, are the triodes in fact paralleled? Your measured plate voltage is 188/220v, which would indicate a nearly dead 5965 - that's is why I am checking.

Paul Joppa


Offline pboser

  • Full Member
  • ***
    • Posts: 118
Reply #9 on: May 16, 2020, 01:27:42 PM
Yes, you guessed right -  line voltage is 131.

I checked the cathode resistors.  Both measure right about 173 ohms.

Also, I forgot to include the thought that the 6CG7 might be the best swap for the driver tube - if I'm correct, it has the same pinout as the 5965?

Thank you!


Peter Boser


Offline Doc B.

  • Administrator
  • Hero Member
  • *****
    • Posts: 9540
    • Bottlehead
Reply #10 on: May 16, 2020, 05:09:25 PM
The 131V is kinda crazy high and not particularly compatible with a PGP8.1 that was designed for 60s vintage 115v mains. B+ tends to be high for this kind of circuit to begin with so some care should be taken to get the already marginal filter caps replaced with something that can handle the excess, and you may need to take some extra consideration in scrubbing off some excess voltage before you hit the tubes. Peej can advise you better on that than I.

And, you better check the filament voltage. You may be running 10% high on the 45 fils, which will shorten their life a bit.

Dan "Doc B." Schmalle
President For Life
Bottlehead Corp.


Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5751
Reply #11 on: May 16, 2020, 05:56:13 PM
(I see Doc B posted while I was writing; his points are good ones.)

Wow - 131v is really high! High enough that I have to ask how sure are you that the meter is correct - are the batteries good, the range optimal, etc? And if it is true, do you want to optimize for that voltage, or do something to reduce it and optimize for the standard 120 volts?

I'm asking about the meter because there are a couple things that seem inconsistent to me. One is that PSUD (the power supply modelling software) predicts 475 vDC at A2 with 120vAC in; that would mean 519 vDC with 131vAC input - and you measured 462vDC. The software is usually much more accurate than that. And the 182-ohm resistor must be a 1% part, so 174 ohms is out of spec. Neither difference is impossible, but both are improbable.

Probably a good idea to measure the 7K resistors too - they are likely to be 5% parts and could be off by enough to matter.

I would still like to know the voltage across the driver cathode resistor, to be sure I understand the original current source.

The 6CG7 heater is wired slightly differently than the 5965 but all the other connections are the same - it would be easy to do. I like that change, but it's less historical and will reduce the gain about 6dB. Since it may or may not be necessary, let's postpone that decision.

Paul Joppa


Offline pboser

  • Full Member
  • ***
    • Posts: 118
Reply #12 on: May 17, 2020, 06:03:47 AM
Gentlemen, first let me thank you for your patience.

The high line voltage was a red herring - I used one of those Kill-a-Watt devices, and never considered that it might be inaccurate.

But then I checked the battery voltage on my best multimeter, a Fluke 117, and also on my second best multimeter.  Both were low, so I replaced the batteries.

Today's line voltage, measured on both meters simultaneously after the battery change, was 118.8/118/3, and then 115 minutes later, 119.5/119.4.

So I re-measured the driver cathode resistors, and got: 174.3/174. (I looked at them to make sure - looks to me like they are the 182R 1% they should be.  Photo of one attached.
Voltage across them: 1.73/1.75

And the output tube voltages:
1- 213.1/236.3
2- 475.9/472
3- 168.3/219.6
4- 213/235.8
(So the PSUD prediction is basically on the money now that we know the correct line voltage.)

The "7K" output tube cathode resistors measured over 5% low:
6.63K/6.60K

Let me know what else to check. 

As always, thank you.

Peter Boser


Offline Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5751
Reply #13 on: May 17, 2020, 10:07:39 AM
That's a big help - thanks!

So, from the "7K" resistor I conclude that it is carrying about 33mA, and the "182" says the 5965s are carrying about 10mA. These are reasonable numbers, and validate my use of PSUD2.

Output stage: 

My calculations predict 453v at the plate of a 2A3 carrying 60mA. This is 53 volts higher than the manual, because the PGP8.1 produces more voltage than the deYoung which was in the original Afterglow. The target for a 2A3 is the value in the spec sheet, 250v plate to cathode at 60mA, leaving 203v across the cathode resistor. Now the driver DC current of 10mA goes through the 2A3, leaving 50mA for the cathode resistor. The revised value should be 4060 ohms. The closest 5% value is 3900, and you probably want a 25 watt or greater rating. The bypass cap of 220uF should have a 350v rating, rather than the 200v originally specified.

Driver stage:

We now have 153v at the plate of the driver, not the 100v of the original. From the 5965 data sheet, we can see that a single triode with 3v bias and 150v plate to cathode will carry 5mA, so two in parallel will match the 10mA provided by the current source. That is easily done with a 300-ohm cathode resistor replacing the 182-ohm original. And at the peak AC driving voltage of 45v, the bias margin is still 1.3v - not Buddha's ideal of 2.0v minimum, but my measurements some years ago found that a 1.0v minimum is probably good enough. So you can retain the 5965s as in the original design.

However, your current 5965s are nearly dead - in your current amp, they should be at 100v on the plate, but you are measuring 168.3/219.6v.  Replace them! (Your 2A3s might also be tired; we'll be able to tell from the voltages after you have made these changes.)

The Plan

Well, what I would do anyhow  :^)

* replace each 100uF power supply cap with two 220uF/350v in series and matching resistors as already suggested

* Replace the 7K resistance with 3900 ohms 20+ watts

* replace the cathode bypass with another 220uF/350v cap (for two amps, that's 10 caps total - enough to get a tiny discount!)

* Replace the driver 182 ohm resistor with a 300-ohm  resistor (1/4 watt or greater)

* Get a fresh pair of 5965 driver tubes

Of course, this is a prediction - so check all voltages again after this is done; we might need to adjust a few values ... :^) Theory is wonderfully powerful, but it's still not reality!

Once it is working, I would strongly suggest that you monitor the driver plate voltage (the 2A3 grid, pin 3, is a good place to check). It should be close to 150v. When it goes over 175v, it's time to replace the driver tube. Note that a new tube may have a high voltage at first, and drift down, taking as much as 100 hours of operation to get to a stable value. Once stable, an annual check is probably good enough.

Paul Joppa


Offline pboser

  • Full Member
  • ***
    • Posts: 118
Reply #14 on: May 17, 2020, 12:13:35 PM
This is exciting!  Thank you for the time spent, and for your specific instructions - that's what I need.

One question- the 1M resistors Doc mentioned for the PS caps - is there one on each cap, like the bleeders on the original amp?
And am I correct in assuming that two series caps are preferred to one higher voltage one because the one would be physically large?

Looking forward to looking around for and ordering parts and getting it back together.  Thanks again PJ!

Peter Boser