My wave isn’t square either, in fact yours is beautiful! I have been reading this thread and have learned a lot in fact I found TrueRTA as a result and many other tidbits of information. I resurrected an amp I built 40 years ago from plans I got from the library. It had one 6267, 12AX7 and 2 EL84’s. I removed all the passive parts and replaced them with new quality ones. I then bought a signal generator and a Tektronix scope to analyze the amp as it sounded quite good. I found that the sound fell off a lot starting at 100hz and by 20hz it was flat dead. From 200hz to 50kz it was extremely flat. I had the source trace on the top and the output trace on the bottom. Using a sign wave I could hardly see any voltage difference in output over this range. I also bought TrueRTA and a cheap usb creative labs sound card to also analyze the amp. The results were awful and are displayed below. I have either used TrueRTA incorrectly or the sound card is worthless. If what I have done with TrueRTA is correct then I don’t believe it. I hope some one here can figure out what I have done wrong, or better tell me the steps/settings I need to properly test the amp. Any help would be greatly appreciate. I have also bought a much better sound card hoping that is the problem.
It's nice to hear some one else got some good out of this thread. I learned a lot from it. There are some great people in this fourm.
I don't think your TureRTA trace looks that bad. You are using a different scale and different settings than I used. Are you using a non-inductive resistor as your load? If not that explains the peak at 70hz (speaker resonance) and possibly some of the other peaks. The dramatic swing at 20khz could be your sound card. Check the input Hz response limit of the card.
When I use TrueRTA I simulate 1 watt output (2.38 v) into 8 ohms. And use a voltage divider to drop the voltage down to about .2V so I don't damage my sound card.
I'm just a beginner with TureRTA myself, if you haven't read the Manual you should look through it. There are several settings that should be made to establish a base lines.
If you are trying to use the sound card as your square wave generator forget it. I have tried with several cards and never got any thing that resembled a SW.
One more thing in the old Mullard book "Tube Circuits for Audio Ampliifiers" they say you need at least -20dB NFB on pentodes. When I started this project I was running about -10dB. Increasing it to -20dB really help the performance of the amp. Sweep your amp with 0 NFB then with -20dB. You will see the vast improvement in the response curve. It will also improve your SW. For the final clean up of the SW play with some caps in the NFB circuit. I found 300-500 pf would clean up the ringing.
I don't think your TureRTA trace looks that bad. You are using a different scale and different settings than I used. Are you using a non-inductive resistor as your load? If not that explains the peak at 70hz (speaker resonance) and possibly some of the other peaks. The dramatic swing at 20khz could be your sound card. Check the input Hz response limit of the card.
When I use TrueRTA I simulate 1 watt output (2.38 v) into 8 ohms. And use a voltage divider to drop the voltage down to about .2V so I don't damage my sound card.
I'm just a beginner with TureRTA myself, if you haven't read the Manual you should look through it. There are several settings that should be made to establish a base lines.
If you are trying to use the sound card as your square wave generator forget it. I have tried with several cards and never got any thing that resembled a SW.
One more thing in the old Mullard book "Tube Circuits for Audio Ampliifiers" they say you need at least -20dB NFB on pentodes. When I started this project I was running about -10dB. Increasing it to -20dB really help the performance of the amp. Sweep your amp with 0 NFB then with -20dB. You will see the vast improvement in the response curve. It will also improve your SW. For the final clean up of the SW play with some caps in the NFB circuit. I found 300-500 pf would clean up the ringing.
Things still look pretty strange to me. Can you describe the amplifier topology - global feedback, etc. IMHO it should be a lot flatter than this, except where it starts to roll off monotonically at the low end and at the high end if roll off is present. I'd expect any ultrasonic transformer resonances to be well outside of the sound card's passband if running at 44.1K..
I agree with you it does not look correct. I suspect it's the $39 sound card, and I have a good one coming. The Tektronix oscilliscope shows it completely flat from 200hz to 50k. My tap is on the secondary of the output transformer. I do have a lot of overshoot with a square wave and I suspect that the TrueRTA with it's quick sweep may also not be accurately representing the response.
The feedback is off the speaker secondary. with a 1K5 resistor and a .001uf capacitor in parallel. I don't have a schematic. I have the beginnings of a circuit board you mount the stuff for my new version attached. I manually constructed this from looking at the components. The front end section is attached. Not enough room for the whole thing.
The feedback is off the speaker secondary. with a 1K5 resistor and a .001uf capacitor in parallel. I don't have a schematic. I have the beginnings of a circuit board you mount the stuff for my new version attached. I manually constructed this from looking at the components. The front end section is attached. Not enough room for the whole thing.
The attachment is in word 2003. I would think that word 7 could read it. I don't know how to get the pic any other way. I cannot screen print and paste here and the PCB software only has a DXF save. Tell me how and I'll try it.
You might try using an 8 ohm-incuctive resistor for the load. The last sweep I saw had the characteristics of a speaker sweep using a mic. Notiably the spike at 70hz or so. I know a single restistor is not the best for simulating speaker load but at least you don't see the speakers response in the sweep.
I think you mean 8-ohm non-inductive resistor........
gwalters:
As Troncones mentioned, if you are connected to a speaker when checking the FR that will change things alot due to the inductance of the speaker voice coil. You need to use a load resistor in place of the speaker when testing.
As a simple check of your sound card, have you tested it in loopback mode? i.e. connect the input to the output and run the sweep.
Thanks for the advice on using a resistor. Was not aware of that.
I did connect the sound card in loopback and 1st display was terrible but the software did adjust to a flat result.
I'll stick in a resistor for a load instead of a speaker and retry. Valuable info I'm getting here. Thanks for the feedback.
Probably clipping during the sweep, (or burst if FFT) that is usually the cause of really screwy looking sweeps - are you using a linear sweep or an FFT to do your response measurement? What sort of attenuation do you have between the output of your amp and the input of sound card? (I use as much as 30dB of attenuation depending on what I am doing.) It's a good idea to set the generator output levels at least 3 - 6dB below 0dBFS for good linearity. (I see issues even with really good sound cards at -3dBFS and above)
Also while it is probably too late a lot of those traces on your PCB are really thin, not necessary to make them so small in most cases. It becomes more relevant if you have a full ground pour on your pcb layout due to increased stray capacitance, but it usually pays in less fragile PCB traces and can be managed with careful layout. I try hard to stay above 62 thousandths whenever possible in tube based layouts, and above 20 thousandths on boards with SMDs..
Your pdf post came through just fine..
Last edited:
Thanks for the comments. The generator was let at -30db as I found clipping as you mentioned. It was also set to a sine wave. I'll try the other choices you mentioned. I still don't trust TrueRTA results yet. Mostly because I don't understand his terms. I'm actually a physicist but have made a living as a computer guy but I am not up on the terms of this trade. The amp has a good bit of gain as I can drive it directly with my iPhone or my wife's nano. Rather neat mixing the old technology with current.
On the PCB traces, most are 20 thousands which are supposed to carry up to 700ma. Some are larger as I thought necessary and there is a ground plane as well as a B+ plane that I'm not using. It will be interesting how it turns out. Your comment on capacitance coupling from the ground plane I had not thought about. The PCB has not been made yet, so I can still make some changes. So far the biggest item is the 10 watt resistor as it is so large. A fun project even if it does not pan out.
There was a comment above on increasing a .1uf coupling capacitor to .47uf to help the low frequency fall off. I'm going to try that when I get the new sound card. I've been learning a lot from those who know far more than me. Thanks
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.