Hi Folks,
Does anyone have any references/articles/procedures on methodologically testing the finished product?
Basically, I measure the heater voltages, the B+ rail voltage and the pin voltages and compare to known normal values. I then have a look at the ripple on the B+. Following this, I put a sine wave on the input and see what it looks like at the other end. At the same time, I use an IR thermometer on all of the relevant power resistors just to make sure none are getting too hot. If all is OK, I hook it up to a test speaker and if it makes an passable sound I hook it up to the good speaker/s. If it then sounds fine, the amp is declared finished.
What else should I do? I have a function generator, a digital and analogue scope (with HV probe) but no distortion analyzer. Even if not required, I am interested in testing as much as possible as I have found it helps me learn.
Just completed another build, it works fine but I would like to put it through the equivalent of customs cavity search.
Regards,
Rob
Does anyone have any references/articles/procedures on methodologically testing the finished product?
Basically, I measure the heater voltages, the B+ rail voltage and the pin voltages and compare to known normal values. I then have a look at the ripple on the B+. Following this, I put a sine wave on the input and see what it looks like at the other end. At the same time, I use an IR thermometer on all of the relevant power resistors just to make sure none are getting too hot. If all is OK, I hook it up to a test speaker and if it makes an passable sound I hook it up to the good speaker/s. If it then sounds fine, the amp is declared finished.
What else should I do? I have a function generator, a digital and analogue scope (with HV probe) but no distortion analyzer. Even if not required, I am interested in testing as much as possible as I have found it helps me learn.
Just completed another build, it works fine but I would like to put it through the equivalent of customs cavity search.
Regards,
Rob
Last edited:
G'Day Rob,
I am in exactly the same position as you. Just finished a 6L6 PP amp, monoblock. Need to test everything is OK before building the other monoblock.
There is a thread about an interface for PC measurement by Pete Millet http://www.diyaudio.com/forums/tubes-valves/155405-test-measurement-interface-soundcard.html
The posted results appear to be using 'Audio Tester' software audioTester
PC based analysis may be the most practical, however, I have a Mac laptop that is doing permanent duty as a desktop, and a Asus netbook that is pretty low power and I am guessing has a pretty low performance sound card.
I am also looking for an excellent post by Tubelab regarding setting up a distortion meter, scope and signal source. Should have bookmarked it...
I do much the same as you re testing, but would like to have some idea about the distortion. I have a new digital scope and an old analogue scope, signal generator and several good DMMs.
What did you build?
Regards,
Chris
I am in exactly the same position as you. Just finished a 6L6 PP amp, monoblock. Need to test everything is OK before building the other monoblock.
There is a thread about an interface for PC measurement by Pete Millet http://www.diyaudio.com/forums/tubes-valves/155405-test-measurement-interface-soundcard.html
The posted results appear to be using 'Audio Tester' software audioTester
PC based analysis may be the most practical, however, I have a Mac laptop that is doing permanent duty as a desktop, and a Asus netbook that is pretty low power and I am guessing has a pretty low performance sound card.
I am also looking for an excellent post by Tubelab regarding setting up a distortion meter, scope and signal source. Should have bookmarked it...
I do much the same as you re testing, but would like to have some idea about the distortion. I have a new digital scope and an old analogue scope, signal generator and several good DMMs.
What did you build?
Regards,
Chris
Great question........
I also do a 10Khz square wave input and look for any ringing and how much the leading edge "corner" is rounded at the speaker terminals on the scope; rounded leading corner=high freq roll off. Non-flat square wave tops also indicate LF roll off problems, ie cannot get down below 20hz, etc.
You can also do a sine input freq sweep and determine freq response by looking at the freq roll off at both ends of the spectrum with the scope on the speaker terminals. With my basic scope, I have to convert volts into dB. As you increase frequency on the signal generator, you are looking for the point where the amplitude of the output sine wave starts to decrease. You'll most likely need to keep changing the time base as you increase frequency to continue seeing a reasonable wave.
Sweeping the sine wave generator freq (way) past the audio freqs with the scope on the output will allow visualization of any high frequency oscillations which can manifest as harsh sounding "edgy" output or other stability problems.
I also do a 10Khz square wave input and look for any ringing and how much the leading edge "corner" is rounded at the speaker terminals on the scope; rounded leading corner=high freq roll off. Non-flat square wave tops also indicate LF roll off problems, ie cannot get down below 20hz, etc.
You can also do a sine input freq sweep and determine freq response by looking at the freq roll off at both ends of the spectrum with the scope on the speaker terminals. With my basic scope, I have to convert volts into dB. As you increase frequency on the signal generator, you are looking for the point where the amplitude of the output sine wave starts to decrease. You'll most likely need to keep changing the time base as you increase frequency to continue seeing a reasonable wave.
Sweeping the sine wave generator freq (way) past the audio freqs with the scope on the output will allow visualization of any high frequency oscillations which can manifest as harsh sounding "edgy" output or other stability problems.
Last edited:
Gimp: Can you give us a few more details?....that sounds interesting. So, use two scope channels, one for the input sq wave and one for the output square wave, and align the leading edges together using the same time base for each channel? Then what?
I was thinking one could measure the time from the input transition to the output transition.
When phase shift is 180 degrees this time should be 1/2 the period of the frequency.
For example, if you had 12.5us between the input and output it would imply 180 degree phase difference at 1/(2*12.5us) =40KHz.
Measure the roll off at 40KHz and it would give you an idea of how much gain margine you had.
Added:
Conversly, one could measure the frequency at which gain went to unity, and then measure the delay and calculate phase margin at that frequency.
If gain went to unity at 36KHz, and you measured 12.5us delay through the amp, then it would imply phase margine:
Cycle time at 36 KHz = 27.8us
180 degree time = 13.9us
Delay = 12.5us
Delta = 1.4us
Phase margin = (1.4us/13.9us)*180 degrees = 5.18 degrees.
When phase shift is 180 degrees this time should be 1/2 the period of the frequency.
For example, if you had 12.5us between the input and output it would imply 180 degree phase difference at 1/(2*12.5us) =40KHz.
Measure the roll off at 40KHz and it would give you an idea of how much gain margine you had.
Added:
Conversly, one could measure the frequency at which gain went to unity, and then measure the delay and calculate phase margin at that frequency.
If gain went to unity at 36KHz, and you measured 12.5us delay through the amp, then it would imply phase margine:
Cycle time at 36 KHz = 27.8us
180 degree time = 13.9us
Delay = 12.5us
Delta = 1.4us
Phase margin = (1.4us/13.9us)*180 degrees = 5.18 degrees.
Last edited:
Chris, I hate to confess on the forum, it's my secret shame - a guitar amplifier. This one was a version of the 18W Marshal. The cabinets are fun to build as well.
I thought of another test - power.
Also, this particular amplifier passed all of my tests with flying colours but in the real world, as my son discovered, when the second channel tone control is turn fully up, the amplifier oscillates. I missed it during testing. It's not so bad, I can play around with it for a while and try and sort it out.
Rob
I usually measure:
- B+, bias voltages, heater voltages
- Output voltage with sine voltage applied on input
- THD
Then I hook up a test speaker - and if it doesn't blow - a good speaker. Then I listen for a while.
I'll then probably look at square wave and maybe look at the distortion spectrum. If I didn't have a distortion analyzer or spectrum analyzer, I'd probably just skip those steps and go by the "if it sounds good, the distortion can't be that bad" mantra.
On a sand amp, I'll measure output DC offset also. With the transformer coupled output of a tube amp, there won't be any DC on the output.
~Tom
- B+, bias voltages, heater voltages
- Output voltage with sine voltage applied on input
- THD
Then I hook up a test speaker - and if it doesn't blow - a good speaker. Then I listen for a while.
I'll then probably look at square wave and maybe look at the distortion spectrum. If I didn't have a distortion analyzer or spectrum analyzer, I'd probably just skip those steps and go by the "if it sounds good, the distortion can't be that bad" mantra.
On a sand amp, I'll measure output DC offset also. With the transformer coupled output of a tube amp, there won't be any DC on the output.
~Tom
Rob; shame on you for saying that
All work is noble dude! I never built a hi-fi amp in my life. I came here to learn how. The #1974 circuit is a great one, and more useful than the big amps. Everybody runs through a PA with a sound man these days and bands sound much better than they did when I was a kid.
From my recent folly's, you need to compare and look at several dynamic measurements.
I use square waves a 1 watt into 8ohms. 1kHz, 2kHz, and 10 kHz. Primarily to set feedback compensation for ringing. You can also look for tilt and other resonse issues.
I then Measure THD into 1 watt 8 ohms at 1kHz. Finally using a 4:1 signal 70Hz,6000kHz, I look at the IMD. Again at one watt as the bench mark.
For extra grins, I then set it up for various other power settings and measure the same.
With some of the software (Audio tester comes to mind) I measure freq response, thd vs. freq, phase vs. freq.
The software is much more accurate in measuring phase than using the leading edge of the square wave. You can see the phase response versus the frequency which helps you identify any frequency abnormalities.
I use square waves a 1 watt into 8ohms. 1kHz, 2kHz, and 10 kHz. Primarily to set feedback compensation for ringing. You can also look for tilt and other resonse issues.
I then Measure THD into 1 watt 8 ohms at 1kHz. Finally using a 4:1 signal 70Hz,6000kHz, I look at the IMD. Again at one watt as the bench mark.
For extra grins, I then set it up for various other power settings and measure the same.
With some of the software (Audio tester comes to mind) I measure freq response, thd vs. freq, phase vs. freq.
The software is much more accurate in measuring phase than using the leading edge of the square wave. You can see the phase response versus the frequency which helps you identify any frequency abnormalities.
Audio Tester V3.0
audioTester
For best results you need a high quality audio card that will sample at 96K or 192K. High quality audio cards can give as much as 140dB s/n (or more) which is necessary to see low level harmonics, etc.
With a standard audio card it is still good enough for amplitude and phase plots, FFTs of amp outputs, but not good enough for distortion analysis of gain stages.
audioTester
For best results you need a high quality audio card that will sample at 96K or 192K. High quality audio cards can give as much as 140dB s/n (or more) which is necessary to see low level harmonics, etc.
With a standard audio card it is still good enough for amplitude and phase plots, FFTs of amp outputs, but not good enough for distortion analysis of gain stages.
Why not crank it up to 0.5dB below clipping and then measure the THD & Square stuff ? If youv'e designed the business end correct, then one should be expecting good results.
richy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.