Selmer Concord 10 Rebuild/Conversion Issues

[Phil_S]

He has a separate filter node for the screens.
The tube spec chart for 6aq5 in class AB1 calls for a max of 250 volt on the plate, it also calls for a max of 30 volts of audio drive signal across the grids, is this the drive signal level you have?

I'm not trying to argue here, just trying to make sure we have the facts in order as what I think I see doesn't seem to agree with what you are saying. I am willing to say I got it wrong.

Earlier in the thread, the OP says he's resocketed for EL84's, Va=282 and Vg2=285. (Maybe having screens above plates is part of the problem; don't think so.) This voltage should be OK for EL84's. We have no clear idea if there is 30 volts of drive signal, but give there are only 2 gain stages, I'm having difficulty imagining that would be too much for a pair of EL84's. The inverter has 0 gain. It looks like the power tubes are cathode biased. If he didn't change the bias resistor from 250 to 130 (consistent with the tube change) nothing bad would happen, just cold bias.

I don't see a separate supply node for the screens. The way I read the schematic, the node labeled OR feeds, in this order, screens, tremolo, cathodyne inverter, and 2nd triode V1. I don't see any decoupling at all on the OR line. (Curiously, the node marked RT feeds only the first triode after the input jack.) My guess is that any time the tremolo wiggles, so does the screen voltage.

[BungleSim]

Phil_S, I think Stevem was just pointing out that in my drawn out schematic I had separated the nodes but you didn't notice it. It's all good! Also, I'm way more of an amateur at this than you are, so do not apologize! I appreciate all input and help, no matter how seemingly basic. The solution to this problem will be very simple, I'm sure.

The real problem in all of this is the amp was clearly modified by another tech before me so I'm dealing with rebuilding a modified amp. He must have converted it to EL84s and no, he did not change the cathode bias resistor from 250 ohms to 130 ohms. Would you recommend I do that for further stability?

Also, I removed a ton of hot glue from the PS filter caps and was able to trace the circuit more accurately. It turns out B+1 is going to the center tap of the primary of the OT and the screens of the power tubes. B+2 is going to the inverter and tremolo circuit, and B+3 is going to the preamp. The three filter caps that replaced the multi-can cap are all 40uF/500V. So, from the rectifier I have a 47uF cap that I added, then there's a 250R resistor and a 40uF cap creating B+1, 1.5k/40uF creating B+2, and 250R/40uF creating B+3. So I think I've got good separation of the power supplies unless you guys think it can be improved.

I started looking at the plate resistors and noticed some were way out of spec (210k measuring over 300k) so I'd like to replace those and see where I stand. Honestly, I'm going to check every resistor and see if they're within spec and replace as necessary.

[Stevem]

You should really add another node of filtering for the screens as when driving the amp and the B+ gets pulled down the screen voltage will change a ton and as such so will the bias which is really what the screen voltage does!

[Phil_S]

All's good, getting on the same page. I think the only thing Steve and I didn't see eye to eye on is how the amp is presently configured. You've supplied the details.

If you are using EZ80 as rectifier, it should not have more than 50uf on it, per the tube spec. I'm still not quite clear what you did. If you added 40uf in series, that works out fine. If parallel, I think you have too much for that tube.

Also, adding a cap where you already have one suggests that you might as well eliminate the old cap as it probably wasn't working. 47uf ought to be plenty of filtering.

Edit: didn't see your question about the bias resistor. The standard for a pair of EL84 is a common 130 ohm 5W resistor. 250 ohm will bias on the cold side. If you don't have 130, anything in the 15-150 range should do. It won't have an affect on stability, but it might improve tone and should increase output. This something that should be done, but doubtful it is related to your problem.

[BungleSim]

Alright, so I worked on the amp quite a bit more on Saturday. It appears the volume swell that is happening is related to the tremolo circuit, though I'm not sure how just yet. As I increase the rate of the tremolo, the volume swell increases as well. I tried removing/isolating the tremolo circuit by disconnecting the plate resistor from that tube stage, but obviously this was a bad idea as I immediately started blowing fuses and I believe the rectifier is now kaput. Once I get a new rectifier, I need to figure out how/why the tremolo circuit is affecting the output the way it is. I wonder if the footswitch is making partial contact when it's supposed to be disengaged.

As for the power supply, attached is a schematic of the current configuration. The three 40uF/500V caps were already present and I added the 47uF/450V to remove the 24V of ripple I was seeing. I am going to bump that back down to 22uF as the 47uF did the job no better. I configured the amp such that the preamp stages were being powered off B+3, the inverter and tremolo were being powered off B+2, and the power tubes' screens as well as the center tap of the primary on the OT were connected to B+1. Connecting things in this way made no audible difference with regard to the problem and at one point I connected the CT of the OT directly to the 47uF cap off the rectifier (we can call that B+0 I suppose) with no improvement.

The tech that converted the amp to EL84s did change the bias resistor to 125 ohms. He also added an electrolytic in parallel with the resistor, but the exact value escapes me at the moment. (220uF seems to be sticking out in my memory.)

[Phil_S]

I think if I wanted to totally isolate the tremolo from the circuit, I'd disconnect the depth pot wiper to the junction of the 220K grid leak resistors. When you disconnect the wiper on the depth pot, there is no ground reference for the pair of 220K grid leak resistors. Where depth pot connected to the junction of the grid leaks, run a temporary ground wire. (Remember, set the footswitch to the off position, too.)

This leaves the tube "on" but the tremolo disconnected.

If you are still seeing voltage sucking from the tremolo tube, then lift the connections to pin 6 to the two coupling caps (.03 and .25). Leave the 210K plate load resistor connected.

The added bypass cap on the power tube bias resistor is not creating the problem.

Adding that cap at B0 is too much capacitance for the rectifier. The tube spec limits it to 50uf. Get a decent quality 20-25uf and it should be more than enough filtering. It shouldn't cost very much.

[BungleSim]

Alright, I've got a working rectifier now so I'm back in business. I've removed the tremolo circuit completely by disconnecting the Depth pot and creating a temporary ground off the 220k's. I actually removed the footswitch completely as that's also not functioning properly and another thing to fix.

My problem now is that it seems that the power cannot stay up and gets pulled down as you strum the guitar. It does not get pulled down only at the cleanest (lowest volume) settings. I have tried two power setups:

SETUP 1
B+0 (right off of the rectifier) is a 22uF cap. The rest is the same as the schematic in my previous post. I realized that the schematic I got off the internet is wrong, though, in the resistor values. So the power setup is... 22uF --> 250R in series with 40uF to ground creating B+1 --> 22k in series with 40uF to ground creating B+2 --> 68k in series with 40uF to ground creating B+3. B+1 feeds the power tubes and the center tap of the OT. B+2 feeds the inverter and (disconnected) tremolo tube. B+3 feeds the preamp tube.

SETUP 2
B+0 (off the rectifier) is a 22uF cap.
B+1 = 250R in series with 22uF cap to ground = 308V ... This connects to the CT of the OT
B+2 = 1k in series 22uF to ground = 297V ... This feeds the power tubes
B+3 = 22k in series with 40uF to ground = 233V ... This feeds the inverter/tremolo
B+4 = 68k in series with 40uF to ground = 157V ... This feeds the preamp

No matter what, playing pulls the B+ down. At max volume and banging hard, I can pull B+ down 30V. Any ideas?

Something I should also mention is the PT HT has a bias tap that I have left floating at the moment. Could/Should I use this at all?

[Phil_S]

Not sure if I understand.
There is 250R between plate and screen?
Then 1K between screen and V2?
It may be that 1K does not provide adequate decoupling. Try something larger, 2.5K - 5K, something in that range. See what happens.

[Stevem]

so at that first 22 uf filter after the rector you play anything thru the amp and the B+ drops down to 30 VDC?
If that's the case then that 250 ohm resistor I would say has a issue, and I do not see why you need that resistor there in the first place as that first filter node should feed the OT center tap and the next node should do the screens of the outputs.

[BungleSim]

I'm hoping these schematics clear things up. The schematic on top details Setup 2 that I described in my last post. The schematic on the bottom is another setup I tried that did not yield positive results.

To be clear, the B+ is dropping by 30V, not down to 30V. So if it's at about 308V then it's getting down to 275V or so when I strum hard.

Lastly, I am not against completely rewiring this power supply. Maintaining authenticity of the original design is not paramount.

[Phil_S]

Thanks for showing those schematics. FWIW, I'm in agreement with Stevem that the 250R doesn't seem to have any purpose and can be eliminated. I think I'd look to use your second schematic. I noted before that the 1K may not provide enough decoupling. If that's true (not sure about it) certainly 250R is not enough. I'd try 5K instead of 250R, which most certainly will adequately decouple the stages. If you don't like the result, you can reduce the value, but if it solves the 30V problem, then at least you have an answer. To be clear, I'm saying use the lower drawing, but change the 250 to 5K.

We've been assuming the 30V problem is somehow related to the circuit, and in particular the tremolo. Maybe we should look at this another way. Maybe the PT is underrated for the circuit. Let's assume you need 120mA and the PT is rated for maybe 80mA (or is somehow partially shorted). If it simply doesn't have enough capacity, it's going to sag like crazy. We've had a change of rectifier tube, so I'd be inclined to rule that out as the problem.

If changing the B+ ladder isn't curing the problem and taking the tremolo off line doesn't cure the problem, we need to look elsewhere. I'd be ready to test the PT to see if it's up to the job. The test isn't necessarily definitive and there is risk of letting the smoke out. This is a last resort. Here's how I do this.

Pull all the tubes so the filament winding is effectively disconnected. Disconnect the high voltage at B+0 so that the rectifier is still operative. Pull the rectifier and tack in silicon diodes in place of the tube. In other words, make it so the PT secondary is isolated from the circuit, but with the first node of the power supply built and voltage rectified. Get several large wirewound resistors (10W, 15W, 25W) of appropriate ohms to simulate the circuit load. Use Ohms Law to figure what you need. Let's assume unloaded B+ is 320 and the amp load is supposed to be 120mA . V=I*R; R=320/.120 = 2666. Approximately 2.7K is the load equivalence of the circuit. Hang this 2.7K load on the B+ supply and measure the voltage drop across the load to determine the actual mA. If the PT stands up to it and doesn't smoke, then it has capacity.

Some cautionary notes...the load resistor gets very hot. 320V @ .012A = 38W, so you need 40W rated load. Two 25W ~6K resistors in parallel will provide for this. When I do this with an unknown PT, I start with a value well above the target value. 10K load is probably an appropriate starting point and work down in increments of 1K or so using a combination of series/parallel resistors to keep the wattage rating appropriate Higher Ohms load needs less wattage. As you reduce the dummy load Ohms, more mA will be demanded, which improves your chances of smoking the PT. Work your way down towards the target. Once you hit the target for the intended use, the testing ends so you don't smoke the PT. If you test to the point it smokes, you get the ultimate answer plus a door stop. It is best to do this where you have a line fuse and a convenient power switch. You will smell the burn before you see it. At first smell, TURN IT OFF.

When I do this, I screw the PT to a board and build out the power supply on terminal strips. But it can be done leaving it in place if you are careful and if there is room. Having 2 meters available is a plus for doing the work, so you can monitor both B+ and drop across the load at the same time, otherwise you are moving back and forth with a single meter. Monitoring the B+ helps you know what to expect in-circuit.

This is some food for thought.

I can draw a schematic for this tonight if you want.

[BungleSim]

Thanks for all of your helpful info! No schematics are necessary as I followed what you said. Since it's a brand new transformer from Mercury Magnetics, I just sent them an e-mail asking what the max current capability of the HT is. It's odd that they list the voltage and current ratings for each secondary winding but only the voltage on the HT. Probably to prevent others from copying their transformers. With that spec I will measure the inline current draw with my DMM at the output of the rectifier and see what the circuit is pulling.

I will also try a higher value resistor for better decoupling. What is the principle behind choosing these component values? Obviously these RC combos create low-pass filters, but what should I be shooting for? The 250R/40uF combo gives a corner frequency of 15.9Hz and a combination of 5k/40uF brings it down to 0.8Hz. Is there some rule of thumb for what frequency your PSU LPFs should be designed to stay below?

[Phil_S]

Unfortunately, I am a mostly self-taught and forum-taught amateur. I am not comfortable discussing corner frequency, although I do have a rudimentary understanding. What I've learned is that a bigger R between stages assures better decoupling. Here I suggested the 5K because using a WAG I guess it is well above what's needed to decouple the stages. So put in the 5K and see what happens. If it stiffens up and solves the 30V problem, that would be a good hint at what should be done. I didn't imagine you'd leave that 5K as a final choice. If it does nothing for the 30V problem, we've ruled out the decoupling aspect and you can go back to the 1K, which I think is a little low.

I am not sure corner frequency is a relevant discussion for the power supply. My simplistic understanding is that corner frequency is relevant where a band pass filter is employed, as in building a tone control in the preamp section. There is no "tone" running on the B+ ladder that I know of. Mostly I see selection of the R values in the B+ ladder as a way to control plate/screen voltage in the downstream stages from the first node.

I'd be more than OK if someone else reading this will say something about my uninformed commentary on the relevance of corner frequency.

[BungleSim]

No worries. When using LPFs in power supplies you are essentially killing high frequency transient voltages that could come down the line. Either way, the corner frequencies of 16Hz and 1Hz aren't really relevant here as those are so incredibly low that it's become obvious to me that it's not a matter of frequency handling as it is of creating voltages that you want to use in the design. Increasing the resistance will have a few different effects on the operation. Hopefully it stabilizes the power supply. I have a 3k/5W resistor on hand so I'm gonna give that a shot.

I heard back from Mercury Magnetics and they said the HT can output 150mA. If I am pulling that much current then I'm going to start looking for partial shorts within the circuit as I believe the preamp should not be reaching that ceiling.

I'll give an update after I'm able to work on the amp again on Saturday. Thanks again for you help.

[Phil_S]

I can't see any way that circuit will pull 150mA. If this is a new MM PT, I think it is safe to assume that's not the problem. You may have to trace the whole circuit to find a short. You may have an unintended whisker somewhere. Use a magnifying glass and a light of good intensity.

3K should be more than adequate.