Ok. If I slow the capture rate to 0.1ms, 10kHz sine wave looks like this:
and the 10kHz square wave:
just a kinda jumpy sine wave. So what do?
thanks!
and the 10kHz square wave:
just a kinda jumpy sine wave. So what do?
thanks!
If it's oscillating as I suspect, this points up the difficulty of dropping a different output transformer into a feedback design. It would be possible to solve the problem but it also could be that the Edcor just doesn't have the bandwidth to sustain that amount of feedback. A step filter at the input could help, or even just experimenting with a larger grid stopper at the input. You could try 25K or even 50K in place of the 10K and see if that helps. I wouldn't know how to implement a step filter in that design, maybe someone else can suggest something. It would be a resistor and cap in series to ground from the first 12AX7 plate.
Try removing the phase lead cap and see if it settles down. Then you'd have to experiment with different values, starting low and increasing until the 10kHz square wave settles down. It seems that 1000pF is not working well with the Edcor.
Try removing the phase lead cap and see if it settles down. Then you'd have to experiment with different values, starting low and increasing until the 10kHz square wave settles down. It seems that 1000pF is not working well with the Edcor.
Ok, I took the 1000pF cap out. Funny, that's where this thread started. Still the same jumpy sine wave:
playing with the signal generator and 'scope, the square wave seems to give way to a bad sine wave above 4kHz. that's within hearing range and bad, ya?
I have 1800pF and 2200pF caps somewhere but doubt I have anything smaller than 1000pF. guess I could get an assortment for my collection.
the PCB has a trace for the Ground right next to the feedback node; I could nix the feedback loop and connect that 1K resistor to ground. Would that be terrible? kinda wanna give it a listen.
playing with the signal generator and 'scope, the square wave seems to give way to a bad sine wave above 4kHz. that's within hearing range and bad, ya?
I have 1800pF and 2200pF caps somewhere but doubt I have anything smaller than 1000pF. guess I could get an assortment for my collection.
the PCB has a trace for the Ground right next to the feedback node; I could nix the feedback loop and connect that 1K resistor to ground. Would that be terrible? kinda wanna give it a listen.
The jagged (jumping as you say) 10kHz square wave signal from the amplifier that looks like a sine wave is caused by two things:
1. Your amplifier high frequency response is rolled off, and strips the 30kHz and higher frequencies of the 10k square wave, leaving only the 10kHz sine wave that is part of (inside of) all 10k square waves.
Seems to be true, since the amplifier has problems already at 4kHz.
(But be sure your signal source square wave looks square, and does not look like a sine wave, or the signal source high frequency is rolled off; and be sure your sine waves: 4k signal source is the same amplitude as when you set it to 1k).
2. Your scope that is set to 0.1ms, the sampler and or display has about 20 per 0.1ms.
Your amplifier is not doing that, it is the scope.
Set the time to 0.05 or 0.01 ms, and see if the wave shape changes (more 'bumps' per 10k sine wave).
1. Your amplifier high frequency response is rolled off, and strips the 30kHz and higher frequencies of the 10k square wave, leaving only the 10kHz sine wave that is part of (inside of) all 10k square waves.
Seems to be true, since the amplifier has problems already at 4kHz.
(But be sure your signal source square wave looks square, and does not look like a sine wave, or the signal source high frequency is rolled off; and be sure your sine waves: 4k signal source is the same amplitude as when you set it to 1k).
2. Your scope that is set to 0.1ms, the sampler and or display has about 20 per 0.1ms.
Your amplifier is not doing that, it is the scope.
Set the time to 0.05 or 0.01 ms, and see if the wave shape changes (more 'bumps' per 10k sine wave).
The 1k feedback is part of the input tube's self bias.
You can not just ground it at either end.
instead, remove the end that connects to the output transformer secondary tap, and ground that end of the 1k resistor.
But that will not stop the amplifier from rolling off frequencies of 4 kHz higher.
Check the 10k grid stopper at the input tube grid (what if it is 100k or 1M . . . then no high frequencies at all).
Check the 1k grid stopper resistors at the output tube grids (what if it is 100k . . . then no high frequencies at all).
Is your schematic at Post # 1 a complete and accurate schematic?
You can not just ground it at either end.
instead, remove the end that connects to the output transformer secondary tap, and ground that end of the 1k resistor.
But that will not stop the amplifier from rolling off frequencies of 4 kHz higher.
Check the 10k grid stopper at the input tube grid (what if it is 100k or 1M . . . then no high frequencies at all).
Check the 1k grid stopper resistors at the output tube grids (what if it is 100k . . . then no high frequencies at all).
Is your schematic at Post # 1 a complete and accurate schematic?
I could not find the maximum frequency response of your scope; it does list a 1 Meg sample rate.
If the scope and signal sources frequency responses are OK beyond 50kHz, then . . .
There is something fundamentally wrong with your amplifier:
bad part
wrong part value
incorrect wiring
etc.
The amplifier schematic in Post # 1 does not/will not have any problem at 4kHz, 10kHz, etc.
If the scope and signal sources frequency responses are OK beyond 50kHz, then . . .
There is something fundamentally wrong with your amplifier:
bad part
wrong part value
incorrect wiring
etc.
The amplifier schematic in Post # 1 does not/will not have any problem at 4kHz, 10kHz, etc.
Looking back thru the build guide at R-type, they only found success with the Fb loop connected to the 16-ohm tap of the Hammond 1608A. With a similar footprint, I won't have to drill any new holes, but I wish I could make these OTs work 🙄
That's good to hear. The only change I've made to the OG schematic is to put a 100k pot on the front end. The circuit is on a PCB so not much can change, except component values. I did check all the resistors with a DMM during the build. But I'll go thru everything today, try to knock up a compleat schematic and some more pictures. Thanks for helping.
w
w
I checked all the components and everything on the PCBs are correct to the schematic. Here's an updated version of that:
and a shot of the interior:
I found an array of small film caps from 470pF-220pF and put em in, one value on each side for A/B testing. I found no difference in the handling of a 10kHz square wave with any of those values in the circuit. <sigh>
and a shot of the interior:
I found an array of small film caps from 470pF-220pF and put em in, one value on each side for A/B testing. I found no difference in the handling of a 10kHz square wave with any of those values in the circuit. <sigh>
Good News, Everyone! I got my hands on a real signal generator, and it seems I'm in better shape than I thought. Turns out my phone can't generate a good square wave at 10kHz. Hooked it directly to the scope, and it falls apart above 3kHz. So I found this thing I got off Amazon same time I got the 'scope, and got it wired up. It uses jumpers to see the range of frequencies produced, and I'm getting a nice square wave at least up to 40kHz:
This generator module has pots, and I can sweep thru most of the 3k-65k range before the square gives way to a hump wayyy up at the top. So we're in good shape?
Furthermore, see that little squiggle at the front end of every wave? What is that? I decided to put one of those 470pF caps in one channel, and a 2200pF in the other. The 470pF channel still had the squiggle, but the 2200pF side did not!
If anything, it undershot that leading edge a bit. I'm going to try a 1800pF on one channel, see what happens...
This generator module has pots, and I can sweep thru most of the 3k-65k range before the square gives way to a hump wayyy up at the top. So we're in good shape?
Furthermore, see that little squiggle at the front end of every wave? What is that? I decided to put one of those 470pF caps in one channel, and a 2200pF in the other. The 470pF channel still had the squiggle, but the 2200pF side did not!
If anything, it undershot that leading edge a bit. I'm going to try a 1800pF on one channel, see what happens...
I have not come across elevating heaters by wiring them to the cathode bias resistors. I think you will have two routes to ground, so maybe there is some cancellation taking place between the PCBs?
I think it is safest to create a resistance divider on the B+, with the negative part bypassed with an electrolytic, and then you have a stable elevated heater supply, and it cannot interfere with the amplifier.
I think it is safest to create a resistance divider on the B+, with the negative part bypassed with an electrolytic, and then you have a stable elevated heater supply, and it cannot interfere with the amplifier.
But that's the point of the center tap of the heater secondary: a path to ground. The heater current wants to return to its source, the windings. There is no path to ground there for the tube's current thru those windings, the small DC voltage there will not flow thru the heater circuit, it just elevates it.
OldHector,
Perhaps you were looking at an SRPP or Concertina circuits when you recomended to elevate the heaters.
It makes sense with an SRPP, and also makes sense with a Concertina (split load inverter).
For the amplifier of this thread, in the most recent schematic (in Post # 72) . . . grounding the filament winding center tap is the perfect thing to do.
Very rarely is there enough cathode to filament leakage current to cause any hum, or quiescent DC disturbance to this amplifier design (Post # 72's circuit)
If there is a problem, the offending tube needs to be replaced.
If there is L to R channel leakage on this amplifier, it will be from some other cause than filament to cathode leakage.
Where channel to channel separation can be a problem, due to filament to filament coupling:
I built a stereo single ended 45 amplifier that only had room for one AC filament winding.
With 25 Ohm resistors from the filament winding ends; and the other ends of the 25 Ohm resistors connecting to the self bias RC circuit to ground.
I tested the channel separation, all the way from 20Hz to 20kHz.
It was only -40dB. Wow, so many perfectionists were disturbed to hear of such a [bad] separation number.
So, I told them to check their best phono cartridge for the separation curves all the way from 20Hz to 20kHz.
No comment? Silence!
By the way, the many many people who heard my 45 amplifier design never made a comment about the "poor" channel separation.
Perhaps you were looking at an SRPP or Concertina circuits when you recomended to elevate the heaters.
It makes sense with an SRPP, and also makes sense with a Concertina (split load inverter).
For the amplifier of this thread, in the most recent schematic (in Post # 72) . . . grounding the filament winding center tap is the perfect thing to do.
Very rarely is there enough cathode to filament leakage current to cause any hum, or quiescent DC disturbance to this amplifier design (Post # 72's circuit)
If there is a problem, the offending tube needs to be replaced.
If there is L to R channel leakage on this amplifier, it will be from some other cause than filament to cathode leakage.
Where channel to channel separation can be a problem, due to filament to filament coupling:
I built a stereo single ended 45 amplifier that only had room for one AC filament winding.
With 25 Ohm resistors from the filament winding ends; and the other ends of the 25 Ohm resistors connecting to the self bias RC circuit to ground.
I tested the channel separation, all the way from 20Hz to 20kHz.
It was only -40dB. Wow, so many perfectionists were disturbed to hear of such a [bad] separation number.
So, I told them to check their best phono cartridge for the separation curves all the way from 20Hz to 20kHz.
No comment? Silence!
By the way, the many many people who heard my 45 amplifier design never made a comment about the "poor" channel separation.
But OldHector is correct, I have connected the artificial center tap of my heater supplies to the top of the cathode resistor for the power tubes. As mentioned in the Heater Supplies section of valvewizard.com, it's a convenient way to slightly elevate the circuit.
dubadub,
Your photos in Post # 73 shows the scope at 20us.
Is that 20us per division?
1. If it is 20us per division, then:
Top photo: the "ringing" of the top photo is about 10us from crest to crest (100kHz).
Bottom Photo: the rise time is about 7us. That means the amplifier's -3dB bandwidth is about 50kHz.
2. If it is 20us for the whole screen (10 divisions), then:
Top picture 'ringing' is 10kHz
Bottom picture bandwidth is about 5kHz.
By the way, for the top photo, I estimate the rise time at about 5us, which gives a -3dB bandwidth of about 70kHz.
Considering your amplifier schematic, and the Edcor output transformer . . .
I expect the amplifier performance is more like # 1 above.
Note: My one ear, and my speakers, do not work at 50kHz, so I would not be worried if I had to listen to your amplifier.
Have Fun Listening!
Your photos in Post # 73 shows the scope at 20us.
Is that 20us per division?
1. If it is 20us per division, then:
Top photo: the "ringing" of the top photo is about 10us from crest to crest (100kHz).
Bottom Photo: the rise time is about 7us. That means the amplifier's -3dB bandwidth is about 50kHz.
2. If it is 20us for the whole screen (10 divisions), then:
Top picture 'ringing' is 10kHz
Bottom picture bandwidth is about 5kHz.
By the way, for the top photo, I estimate the rise time at about 5us, which gives a -3dB bandwidth of about 70kHz.
Considering your amplifier schematic, and the Edcor output transformer . . .
I expect the amplifier performance is more like # 1 above.
Note: My one ear, and my speakers, do not work at 50kHz, so I would not be worried if I had to listen to your amplifier.
Have Fun Listening!
That's more like it! Good work. The "squiggles" at the leading edge are called ringing, and they indicate an uneven frequency response, of ten above the audible range. The goal is a perfectly flat top on the 10K square wave. Some people like a little bump at the leading edge, but the rest of the top should be flat. You're second shot rounds the leading edge, so yes, a small phase-lead cap is in order.
Now you'll want to test a bit for stability, and you do that by dabbing a .047uF cap across the outputs and see if the square wave holds up at all. If it collapses the amp is not very stable, which is something you may want to address. If it hangs in there, try increasing the capacitance to .1uF, and at most .22uF.
Another good test is to disconnect the load and see if the square wave hangs in there. Then, ultimately, you can try a purely capacitave load, like .001uF on up to, say, .047uF. This is a tough test but it will tell you a lot.
It's good to have a variety of small caps handy for testing like this, say, .001, .047, .1 and .22uF.
This cheap 'scope came with no documentation, but it's fair to assume it's /per division, and your assumptions have held up so far.
That little sig gen shat the bed after those last tests, but I'm listening now with 1800pF caps in the Fb loops and sounds pretty good! 2200pF was a little weird, harsh highs n boomy bass. I'll put 1200+1500pF caps in my next Mouser order and see how they do.
Now to see about a nice diy signal generator 😀
Thanks guys!
Will
That little sig gen shat the bed after those last tests, but I'm listening now with 1800pF caps in the Fb loops and sounds pretty good! 2200pF was a little weird, harsh highs n boomy bass. I'll put 1200+1500pF caps in my next Mouser order and see how they do.
Now to see about a nice diy signal generator 😀
Thanks guys!
Will