More Amp Construction

@tubegopher stopped by a while ago and made me a deal I couldn't refuse on some 5K Edcor transformers, and I decided to whip up a quick breadboard today so I can get them built into something and off the bench.
View attachment 41402
Bonus points to whoever can identify the tubes used other than the 6BL7.

I wanted to bread board this to be sure it worked properly since the datasheet for the driver tube was only available in a different language, and plate curves for the output tube are sort of non existent. On my 550V bench power supply, this hums along just fine making 25W, and with the power supply I have in mind I should see about 30W. The Edcor output transformers will do 20W at 35Hz and 25W at 50Hz, so there will be a little power limiting down there, but overall I think this will be a fun project.
I just saw this post and I am glad to see you moving ahead with this amp. I will leave the guessing to others that have a better understanding than me on these type of tubes. It is over my head!
 
Hello, new member here. Paul, I haven't yet read this entire thread, so I apologize if what I'm about to ask has been discussed previously. I'm working on a pair of 211 monoblocks that I built some years ago, then put into storage to deal with other responsibilities. Actually, I can't believe how long it's been, and how much my construction techniques have changed in the intervening years. Anyway, I see you have also worked with transmitting triodes, and I'm wondering if you have ever used RF to power the filaments. I have a method for accomplishing this, but it's one of the areas of the previous work that I'm revisiting.

Jack
 
No, I have not done ultrasonic AC heating. When I last looked into what was being done, I saw a fair number of people using ~44kHz AC to heat DHT filaments.

With the 211, generally you don't have a lot of bias voltage developed between filament and ground, so using an off the shelf SMPS line to DC supply ends up being pretty convenient.
 
Paul, let me know if you prefer this discussion to be moved elsewhere. I don’t mean to derail your thread. Incidentally, I’ve had a chance to read a few more pages, and your work is very impressive.

I designed and built the original prototype for these monoblocks more than 15 years ago. It was the first amplifier I ever built that used a DHT output. After experimenting with several methods to cancel 60 Hz hum in the output stage, I discovered that the 211 heater creates significant harmonic energy that can’t be easily removed. This energy isn’t exactly in phase with the fundamental, so I theorize that it’s due to the very long filament structure vibrating mechanically as it heats and cools at the 60 Hz rate.

Anyway, when I used DC to eliminate the hum, the music seemed to lose some of its “life” and was less engaging. I tried several types of DC supplies, including switching power supplies that offer considerable common mode isolation and should be “inert,” but nothing helped. Eventually, I found an inexpensive means of generating an ultrasonic square wave at roughly 65 kHz, and that resolved the issue. I published the concept on the audio forum where I was active at the time, and several individuals later adopted the technique.

Although I remember this method being sonically transparent, heating with a sinusoidal wave would make me more comfortable. Simple filtering might be an option, but I burned up one of the supplies in an early attempt to do that. So, this is one of the areas I’m revisiting in an attempt to update these two amps as I reassemble and test. I recently also devised a less painful means of impedance selection for the Magnequest OPTS, and a way to stabilize the output bias. These are direct coupled designs using a choke-loaded 6BL7 CF, and the driver tube drifts over the first 10-20 minutes of operation. It’s not enough to damage anything, but it really annoys me. :)

I see you frequently use the Hammond 156C 150H choke for loading CFs. Have there been any “gotchas” using this approach? The low impedance at the cathode would seem to make choke characteristics (resonance and such) much less an issue.

Here’s a pic of one of the amps in its current state. Unfortunately, I’ll need to completely disassemble the chassis and perform more metalwork (such as adding bottom covers) before I can start the reassembly process. LOL, I’m starting to remember why I stored them for so long.

211 Amp 02 sm.JPG


Jack
 
I haven't noticed any issues using the 156C chokes, but at the same time if the grid current, cathode follower current, and bias voltage aren't quite right for the 156C, I will just generate a negative rail for the cathode follower instead. This had the added advantage that the grid of the output tube stays very negatively biased while things warm up, which keeps me happy, but as a downside the whole heater cathode insulation stress sure comes more into play!

When I have listened to the AC vs. DC comparisons on DHT finals, I can understand how some would prefer what they hear with AC heating, but to me it sounds a little less honest than DC, though to a large degree these differences are diminished in my experience with low filament voltage tubes like the 2A3 and 45.

Are you running FS-100s on that chassis? That should be a sweet project!
 
While discussing amp construction. Are there opinions regarding mirror imagining monoblocs vs not doing mirror images?
 
While discussing amp construction. Are there opinions regarding mirror imagining monoblocs vs not doing mirror images?
Mirror imaging requires a bit more work, as you end up doing more layout changes than you think you'd need to in order to build the second amp! You can also run into issues in one amp that you didn't run into with the other.
 
Yes, I'm using 10K FS-100s. They're terrific transformers, and it's very sad that Mike is no longer making any of this.

I understand what you mean about the driver. These have a -80V supply under the chokes. Also, these amps move into A2 when output exceeds about 15W, so the driver has its own B+ supply and transformer. Both sections of the 6BL7 are in parallel, and this draws quite a bit more than the 8 mA rating of the Hammond choke. I was mostly interested to hear how this was working for you, being as it's essentially a PS choke. Considering the low operating impedance of the circuit at this point, It makes sense to me that nothing too special is needed, and I think it's a smart thing to do. Simulations indicate only 30-40H is needed in my particular circuit, and this is one other area I haven't finalized.

That's funny what you said about the "honesty" of DC heaters. One person I ran across years ago thought the difference was actually due to 60 Hz sidebands (IMD) riding on everything. I can see how that might happen, and how it might even have a seemingly beneficial effect, somewhat like the musical characterization of 2nd harmonic energy. However, I switched back and forth multiple times between DC and ultrasonic heating (which doesn't create such sidebands in the audible spectrum), and the effect seemed consistent. Not trying to "sell" you on this idea, it's just something I need to pursue.

Some of your construction methods for the large amps are very similar to mine. These two are constructed mostly from 1/4" 6061 plate. The front panel is 1/2" thick, which allows blind mounting without the need to weld tabs onto the back side. I do all the work myself, so I have to look for ways to accomplish what I want without accumulating too much expensive machinery (like a TIG welder).

Thanks for your comments on all this. I'll try to read through the remainder of the thread over the next few days.

Jack
 
Yeah, if you want to run more current then there are certainly other chokes. We actually sell a 40H/25mA plate choke that would work well for this purpose if you wanted to run more standing current. I tend to use the extra half of the BL/BX7 as a series regulator for the cathode follower triode.

The sidebands are certainly real and audible. The tube itself will also not have one end of the filament at a different bias voltage than the other with AC being used, and that could also contribute.
 
Thanks for mentioning the chokes, I wasn't aware Bottlehead stocked that value. I have two types on the shelf that might work, but the 40H might be preferable, depending on DCR. With -80V under the choke and roughly -50 on the grid of the 211, the 30V drop and DCR of the choke determine 6BL7 current and dissipation. The 6BL7 operates with 360V on the anodes.

There's a lot involved in this upgrade overall. I decided to add an auto-bias circuit to stabilize output stage current, and that turned out to be more complex than I originally thought. It's one of the complications of A2 that output stage cathode current can't be used to monitor bias for this purpose, because it changes at higher power. Anyway, the circuit is complete now on paper. I'll just have to add a PC board and low voltage +/- supply for the opamp that controls bias.

Guess I need to get started on the metal work if there's a chance of getting this done by spring. :)

Jack
 
I agree, rarely have I made A2 amps that will self bias reliably. The 811A amps I make do it well enough to be left alone, with some adjustability in being able to use different cold cathode regulator tubes before the DC cathode follower do move around the bias voltage a little bit.

For what I do, especially since some of this gets shipped, I just use meters and an adjustment pot. The op-amp solution will certainly work, but I'm always wary of tube amp hiccups taking those things out.

The full wind impedance of the 40H choke is about 800 ohms.
 
Thank you for the info on the choke. I need to measure the inductance of one of the types I have with DC flowing through it, then I'll make a decision.

About the bias, output current gradually drifts downward in this design over an initial period of 20 minutes. Maybe it was even longer, I don't remember now. If it's set to 80 mA cold, the drift ends at 70 mA or so. This doesn't degrade performance significantly, but I'm a perfectionist and it really bugs me.

I've looked at many different means of controlling this, but most are fundamentally thwarted by A2 operation. Given the fact that average cathode current changes when the amp enters A2, only two parameters can be used for electronic monitoring. One is the DC at the grid of the 211. A high impedance connection there driving an integrator can be used to keep it steady. However, stabilizing the voltage isn't sufficient to compensate for aging of the 211. The amps would still require adjustment from time to time. The second means is to monitor average anode current. That doesn't change when the amp is driven into A2 (if it does, the amp isn't linear). This can be accomplished with a small value resistor in series with the B+ negative return. The "top" of the resistor will be positive, carrying a value commensurate with anode current.

Using SPICE, I originally devised a simple means using a single FET to monitor this point and make corrections. However, startup is complicated. The auto-bias initializes first, and because the 211 isn't conducting yet, it pushes grid bias to the positive end of its window. When the 211 eventually conducts, the circuit senses the over-current, but as an integrator with a long time constant, it responds too slowly. This causes the 211 to be way over-current for an extended time, and it self-destructs. The only solution I've found is to use a timer to precondition the auto-bias circuit so it outputs a full negative voltage. The timer would be set to expire after the 211 conducts, at which point the grid will gradually move less negative to reach equilibrium. Unlike the FET appoach, the timer requires a source of low voltage VCC. Once the need to include that additional power supply is accepted, it's easy to replace the FET with an opamp for more refined control. I spent hours on this problem, because the FET is a really simple, elegant solution that doesn't require additional power. I just couldn't devise an equally simple means of preconditioning it during startup.

Yes, tube amps often eat semiconductors, and I share your concern in this regard. The circuit I've come up with will use the opamp to directly drive the grid of a 6SL7. Nothing else will be connected to that point, so unless the tube shorts anode-to-grid, the opamp is completely isolated from high voltage. Also, if the 6SL7 opens, 211 bias will move to -80V, cutting it off. That's the plan, anyway. :)

Jack
 
The 211 itself warms up pretty fast. IIRC, it's only a few seconds. However, actual current flow doesn't occur until the 5R4GYB in the HV supply conducts. That happens slightly later. Overall, I think anode current begins to flow about six or seven seconds after power-on.

Just to be clear, the issue is not so much when the 211 conducts, but rather how fast it goes from zero to full conduction. This happens quickly when the rectifier turns on, and that's the "ramp" that the auto-bias circuit would have to keep up with. Because of this, auto-bias has to be preconditioned to output maximum grid bias, then released to do its thing sometime after the output stage is ready to conduct.

In practice, this can be accomplished by applying an external positive voltage to the integrator components (RC) at the input of the auto-bias circuit during the warmup period, similar to the positive voltage that appears on the B+ return resistor when the 211 is conducting. My circuit on paper uses a 10 ohm return resistor, so roughly +0.75 V or +0.8 V will be input to the auto-bias under normal operating conditions.

I can upload the schematic(s) for the auto-bias later today if you're interested in seeing the work.

Jack
 
Anode voltage on the 211 is above 1KV. Even if the 5AR4 could handle it, all these rectifiers turn on faster than the auto-bias can correct. The circuit has a very long time constant to prevent interaction with the audio signal.

Here's the opamp version of the auto-bias that I came up with. The FET version probably isn't worth pursuing at this point. The preliminary values in this circuit force the auto-bias to apply full negative voltage to the driver for approximately 10 seconds. After that, the timer expires and auto-bias output gradually turns on the 211, eventually reaching equilibrium. R1 or R2 would be made semi-adjustable to adjust bias current to the desired value.

EDIT: In my haste to transfer this from the SPICE schematic, I reversed the inputs to the opamp. Input from the 10 ohm rsense resistor should be connected to the inverting input of the chip. The non-inverting input should. be grounded.

Jack

Auto-Bias-Opamp.jpg
 
Last edited:
Back
Top