New Tube Amp Tests?

Ideally you want the tops and bottoms of the 10Khz square wave to look like that in post #7. Depending on the make of your output transformers you may or may not be able to achieve that. One way you can dial it in is to put R13/R14 on a pot temporarily instead of a fixed resistor. Run the square wave through it and adjust the pot until you get the square wave tops to look as much like post #7 as you can. Measure the pot and then install the closest size fixed resistor of that value. This is somewhat easy to do for a well wound transformer, and quite a bit more difficult for one that exhibits more winding resonances or has other anomalies.

Also, because speakers represent complex loads, you need to test HF stability under different types of loading conditions. You want a stable amp under any type of speaker load you can throw at it (because you certainly don't want your amp to exhibit instability at certain frequencies when actual speakers are connected and playing). You would do these tests at 1 watt output:
  • No speaker load
  • Resistor only load representing the output tap impedance selected (ex: 8 ohm resistor on the 8 ohm tap)
  • Paralleled cap with resistor load (varying sizes of capacitor from say .1 uF to 16 uF or so)
  • Cap only load (varying cap sizes from .01 uF to 16 uF)
If the amp does go unstable under any of those loading conditions, you need to go back to your feedback/HF tuning and./or basic design, and retweak things until you get things stable. Again, this is all highly dependent on the output transformers.

Of course, and this is very important, measurements only tell some of the story. You need to listen to it for a while too and see if you like it. It usually takes me a week or so to really get a feel for how the amp sounds.
 
I tried changing C3/C4 up/down in value but that didn't improve anything.
You say it didn't improve anything,if it didn't have any effect at all,your fb loop may not be hooked up.
Does the FB terminal(NF) connect to the 16 ohm tap on the output transformer?Is the transformer COM terminal connected to ground?
 
Depending on the make of your output transformers you may or may not be able to achieve that. One way you can dial it in is to put R13/R14 on a pot temporarily instead of a fixed resistor.

That's what I did! Only I used a resistor box instead of a pot.

Of course, and this is very important, measurements only tell some of the story. You need to listen to it for a while too and see if you like it. It usually takes me a week or so to really get a feel for how the amp sounds.

Will do!

You say it didn't improve anything,if it didn't have any effect at all,your fb loop may not be hooked up.
Does the FB terminal(NF) connect to the 16 ohm tap on the output transformer?Is the transformer COM terminal connected to ground?

It was definitely hooked up. Changing the feedback cap value made things worse or the same, but not better.
 
I just realized all my power measurements probably aren't valid because I only injected a signal into one channel at a time. The load from the other channel will create sag in the power supply which will affect the channel I'm measuring. But I only have one function generator. Can I hook the two channel inputs in parallel? Or should do this through some resistance (IE, connect the generator to 2 resistors, one going to each channel input)?
 
they are totally valid and the rite way to do it. One ch at a time. the other ch wil not produce power as it has no signal to amplify.
Testing both chs together for a short period is ok to do but puts a strain on some components like transformers
so be gentle with your gear and dont do that very often. the untested ch should be shorted to gnd so it wont start
oscillating.
 
Ok, thanks. If I wanted to do it briefly just to find out what my max current draw is, could I hook the two channel inputs in parallel to drive them from my func gen?
 
Awesome, thanks. Sorry one more question! I measured 12.4 Vrms at the output of my amp before it started to clip. Since I was using 8 ohm load resistors, does it follow that I have an approximately 19WPC amplifier?
 
That's correct

Thanks!

lso, because speakers represent complex loads, you need to test HF stability under different types of loading conditions. You want a stable amp under any type of speaker load you can throw at it (because you certainly don't want your amp to exhibit instability at certain frequencies when actual speakers are connected and playing). You would do these tests at 1 watt output:

Sorry, I forgot to follow up on this. Can I use ceramics for this purpose? I assume I should try this at 100 / 1K / 10K Hz frequencies? And what am I looking for on the output? Ringing?
 
I've always used film caps. Not sure how much ringing ceramic caps will contribute. Test from 10 Hz to at least 250KHz. When testing the output for HF stability, you are looking for sustained oscillation. If you have a scope, this is really easy to see--the trace will go from sane (i.e., a sine wave or square wave) to something totally unrecognizable, or it will be a sine-ish looking wave but at much higher frequency or a combination of several frequencies. Also, sometimes you can "hear" the instability directly as mechanical noise from the output transformers. Sometimes the oscillation only occurs on certain portions of the wave, such as at the crests or on the upswing or down swing. It's hard to describe, they sorta look like little cocoons lodged on the sides of the signal. Here a hand drawing of what I mean:

Scan.jpg
 
Cool; thank you so much! Forgot to ask: I was told not to run a tube amp with no load. Is that bogus?
 
At 1 watt output under an appropriate dummy load representing the secondary load impedance, it's okay to disconnect the load for this test. Definitely don't want to run a tube amp at full power without load!
 
Just for my own comfort, I never run a tube amp with no load R attached.
But thats just me.
with no signal in its probably safe to do.
 
Well, the only way to check a tube amp for stability under no-load conditions is to operate it with No Load. I make these tests as quickly as possible, because it's pretty common for developmental circuits to take off like a big bird, i.e. oscillate at full amplitude. I haven't blasted any tubes or OPTs doing this yet, but it's just a matter of time.
 
It's only happened to me once--lost an output tube when the resistive load somehow came disconnected from the alligator clip and the amp went into sustained oscillation, at full power. I was 10 feet away from the amp when that happened, and I couldn't get over to the amp fast enough (4 or 5 seconds) to shut things down. It blew an output tube then blew the fuse. They were Chinese KT66's so I didn't worry about it too much.

But the lesson I learned was always be within arm's distance with a watchful eye when doing the no load test on your unproven or DIY builds.
 
I've settled on a no-load test tactic that minimizes the risk of damage. My workbench is equipped with a patch panel that switches between speakers and dummy loads with break-before-make relays. I attach a scope probe to amplifier output and switch loads while scrutinizing the scope display. If I see a burst of oscillation, then I know I've got more work to do.
 
I ordered some 0.01, 0.1, 1, 10, and 15 uF capacitors and hooked each one across my 8 ohm load resistor. I put a sine wave on the input that generated a 1 Wrms output. For each cap, I varied frequency from 10Hz to 250KHz. The 1uF and above capacitors exhibited some strange behavior between 30 Khz and 100 KHz. You can see it below. This is the 1uF, which was the worst of the 3:

scope_8.png

scope_4.png

scope_5.png

scope_6.png

scope_7.png


Is this instability that I need to address?

I also tried the no load and cap-only tests. When I turned on the source, the output wave would start rising and falling as if there was a very low frequency AC offset applied to it. The amplitude of this offset would increase quickly so I would shut it off before things got out of hand. Is this also instability that needs to be addressed as well? Offset aside, the actual waveforms looked fine (or, similar to what they did with the load resistor in parallel).
 
Your HF stability looks fine to me. Check all of the other combinations of loads if you haven't already though (different variations of cap sizes in parallel with dummy resistor load).

The other problem you describe of a source signal wobbling up and down is indicative of LF instability, and that definitely needs to be dealt with. What is the apparent rate of wobble? (probably somewhere between 1 and 10 Hz I expect). Anyway, to get to the bottom of that, we'll need to see the full schematic, including power supply, know how much feedback you have applied, and ideally what brand/make of output transformers you are using. Does the wobble stop if you disconnect the feedback?
 
Back
Top Bottom