Squeals of displeasure from DIY Dynaco ST35 clone

[dcgillespie]

If I can add two quick comments to make sure you "know what you think you know":

1. First: Remove the driver tubes and operate the amplifier. Is the amp completely silent in this mode? If so, then the problem is almost surely NFB related. However, if the anti social behavior persists with the driver tubes removed, then the problem must reside elsewhere and proves that no amount of work on the NFB system will resolve the issue, because the problem is not NFB related.

2. If the problem is proven to in fact be NFB related (likely), then consider this: Because you are using non-specified OPTs, then all bets are off that the color code of the primary wires matches those of the original Z-565 transformers. Oh, they may use the same color wires with the same color tracers as well. They may even use the same colors to indicate which are the plate and screen connections as well. In a previous exercise, I had you make sure that the primary leads were all connected properly regarding the plate and their relevant screen connections. However, one other potential issue remains regarding use of the Edcor transformers in place of the Dynacos: Phase.

There is simply no standardization (certainly back in the day) as to what (for example) a blue primary lead represents, relative to a non-black secondary lead. The blue primary lead certainly implies a plate lead, and a non-black secondary lead implies an impedance tap as opposed to the Common connection, but there is no guarantee that the relative phase between these two leads is the same between two transformers from different manufacturers.

Therefore, as one last test before digging into the NFB circuits, disconnect the coupling caps on the end that connects to the output tubes, and then reconnect them in reverse from how they were originally connected. This will reverse the phase (polarity) of the signal appearing at the secondary of the OPT. Use the correct NFB resistor for the secondary tap being used, and turn the amplifier on. If the problem is at least as bad or worse, then the phase was correct as previously connected. But if the problem is resolved or much better, then it indicates that the color code of the OPT leads do not match, and the phase of the push-pull signal needs to be reversed.

Let us know!

Dave

 

[triode17]

OK, I think we need to look at the "SDS" filter board you have. Very often motorboating is caused by insufficient decoupling between the PS filter stages. There may not be high enough resistance between your caps to create the necessary separation of filtering. I'm not an encylopedia of motorboating causes but it's been cured many times here by getting the PS filters right. You said you have 5 stages with 220uf each. That nothing like the ST-35 and I want to say it's so excessive that your low frequency oscillation is likely coming from there.

I'd build a stock(ish) PS just to get a working amp and then try your hand at bumping up the cap sizes. It could bring your voltages back into a more normal range, too.

I'm with you here about the excessive capacitance. Just looked at the original schematic and it only has a 40uF followed by a 20uF! I might double the values to 80uF and 40uF but that's all that needs. You can have too much capacitance.

 

[currituckco]

Looking up videos of "motorboating" on youtube, I think I'm incorrectly calling the noise that - it's not a low frequency growl, more like a more mid-range oscillation. I do believe it must be NFB related, as the amp is quiet with the drivers pulled. I'm going to check the phase now.

Thanks all for the input!

 

[primosounds]

Thanks so much, Dave - very I informative (if disappointing) post. I have much to learn, and I appreciate the knowledge you're imparting! I'm almost tempted to get a scope and try to learn to do the work of tailoring the circuit just for the learning experience, but for now I guess I'm off to try to source some original (or reproduction) Dyna OPTs. Anyone have experience with any worthwhile modern versions tailored to this circuit?

Alternatively, you can use your OPT in a different circuit. I like the EICO HF86 which is a very nice sounding amplifier as well.

 

[gadget73]

There is actually very little difference between the ST-35 design and the HF-86. The only real difference is a step network after the voltage amp and thats all determined by the particulars of the transformer in use. Since he's not using the specific Eico iron those values probably aren't going to be exactly right either, so may as well tune the ST-35 circuit. It may well end up looking closer to the Eico schematic when all is said and done anyway.

but yeah, if the phase ain't right, none of the rest of it has a chance of being proper either. I've done that myself, the specific horrible noise it makes varies but its definitely not a happy one. And it usually has happened to me when trying mystery transformers, wiring them up is a guess as to what its going to do. One amp I actually fixed by reversing the secondaries because it made them more useful values.

 

[currituckco]

Well, it's definitely not the phase! Much worse with it flipped. With the 47k nfb resistor and the correct phase it makes a sort of popping oscillation, I don't know if it's the low frequency growling that is generally called motorboating. This oscillation shows up as the amp is driven harder - I don't hear it with the volume low, just as the source volume increases.

I also decided to bite the bullet and order a scope & function generator. I need to learn how to use those tools anyway.

 

[dcgillespie]

At least that element is determined and now eliminated as a possible contributor to the problem. You might also check the operation of the amplifier with the NFB loops (both of them) disconnected completely, just as a double check of the integrity of the basic amplifier circuit. In that conditions, there should be no misbehavior at all if all is well.

Dave

 

[currituckco]

At least that element is determined and now eliminated as a possible contributor to the problem. You might also check the operation of the amplifier with the NFB loops (both of them) disconnected completely, just as a double check of the integrity of the basic amplifier circuit. In that conditions, there should be no misbehavior at all if all is well.

Dave

No misbehavior with all NFB disconnected. I've finally acquired a scope, and I can see a square wave with no feedback that has considerable overshoot. I connected a 100k pot and a variable capacitor at the spot for the NFB resistor and bypass cap, and when decreasing resistance the amplitude of the square wave decreased without decreasing the overshoot, until instability. When applying capacitance it looked like the wave just thickened with distortion until instability. Any suggestions on how I should go about trying to tailor this for best results?

Thanks much for your help!

 

[Dandy]

Have you attended to the power supply capacitance issue yet? See post #24.

 

[gadget73]

OK, try this while you have the NFB disconnected. At the plate of the 12DW7 voltage amp, try a 2.2K resistor in series with say a 100pf cap to ground. See if the overshoot quits or reduces. Fool with those values as needed until it does. The cap value will shift the frequency at which it starts to roll off, the resistor changes the amount it rolls off.

Once it stops, see if the oscillation problems continue with the feedback connected.

 

[currituckco]

Have you attended to the power supply capacitance issue yet? See post #24.

I don't believe the power supply capacitance is causing any problems - especially since the amp doesn't oscillate or misbehave with feedback removed. So for now I think I may leave the power supply as it is...correct me if I'm wrong, of course!

 

[currituckco]

OK, try this while you have the NFB disconnected. At the plate of the 12DW7 voltage amp, try a 2.2K resistor in series with say a 100pf cap to ground. See if the overshoot quits or reduces. Fool with those values as needed until it does. The cap value will shift the frequency at which it starts to roll off, the resistor changes the amount it rolls off.

Once it stops, see if the oscillation problems continue with the feedback connected.

I'll try that this afternoon, thanks!

 

[GordonW]

Looking up videos of "motorboating" on youtube, I think I'm incorrectly calling the noise that - it's not a low frequency growl, more like a more mid-range oscillation. I do believe it must be NFB related, as the amp is quiet with the drivers pulled. I'm going to check the phase now.

Thanks all for the input!

IME, midrange oscillation (in the 150-400Hz range or so), oftentimes is a phase issue.

Dave Gillespie helped me conquer that on one of my amps, with some surplus military output transformers- we had to go with a step network (series RC in parallel with the plate resistor) on the plate of the gain tube of the output stage, a phase cap on the feedback resistor, AND a Zobel across the output terminals to get it completely stable.

I can't remember what order we did this- I THINK we started with the RC network across the plate resistor of the gain stage, to get it as good as it could be with just that, then added the cap across the feedback resistor, and once we did that to best effect, then added the Zobel, in that order.

With an unknown transformer characteristic, it can require starting from the very beginning of stability tuning, like that...

The good thing? Once we got done with all that tuning, the amp sounded AMAZINGLY good. Much more open, airy and effortless sounding, than it was before we got everything right...

Regards,
Gordon.

 

[currituckco]

It sounds like what gadget73 is suggesting is more or less step one in your procedure, no? So this sounds like a great starting point for me.

Thanks, guys!

 

[currituckco]

Okay, here are some pics...

This is the square wave with no FB and no compensation. I think I was incorrect in describing the issue as overshoot:

This is with 2.7kR and 33pf from the VA to ground. It actually creates some undershoot, made worse by increasing capacitance. I saw no difference from changing the resistance:

This is putting C6 back in. This is the 18pf from the V6 UL tap to the VA cathode:

This is additionally adding 100k feedback resistor:

And just for reference, I tried to get the square wave before the amplifier, which for some reason I had trouble getting (inexperience is the reason). I am using a android phone app to create the square wave, so I'm not surprised if it is a, shall we say, suboptimal situation. I bought a cheap function generator but it is non-functional, so I worked with what I could find:

 

[s-petersen]

On the phone pic, adjust the scope to the next higher sensitivity range to see your input better, with the tilt on your square waves something looks wrong at 1K. Is the common lead of the opt grounded? Do you have a 16 ohm load resistor connected? You should be using a higher signal input and output, you want about 5v at the output at the 8 ohm tap
I didn't see it in the text, I may have missed it, Is the driver board a SCA35 also?
try removing C3 from the board. C3, C6 and C7 are all transformer dependent C1, C4 and C5
might need to be increased to flatten the tilt, Dave would have a better idea than I for sure.

 

[Dandy]

I don't believe the power supply capacitance is causing any problems - especially since the amp doesn't oscillate or misbehave with feedback removed.

That's good. I was just checking whether no misbehavior meant stability without NFB.

 

[gadget73]

What frequency is that square wave?

Another question, what does it look like when you connect the sig gen straight to the scope? I'm just curious what we're starting with. Unless its a good sharp square going in, there is no chance of getting one on the output.

Looks like there is some hum or something in there too, unless thats just coming in from the sig gen.

 

[s-petersen]

According to the scope 1K (I was going to ask also)

 

[currituckco]

I downloaded a bunch of mac function generators on my laptop, but all of their square waves had really bad ringing straight into the scope. But I realized my scope has a very flat 1k square wave to use to calibrate the probes, so I tried that through the amp. Here's what I got.

This is the amp with no FB:

This is with 22k feedback resistance added:

This is with the 22kR and additionally the 18pf feedback cap from the UL tap:

The other feedback cap (in parallel to the resistor) caused oscillation.

This is if the FB resistor is increased (and no FB caps):

Or decreased:

It definitely looks like I have a problem with low frequencies, right?