I was thinking about starting to use Tractor to make it easier to run tests on amplifiers, but want to have the actual data and not a “Pass/Fail” result. For example, on a power amp, have the operator (me) set it up for say 5watts into 8 ohms at 1Khz, take snr, thd & gain data, and then take the input level required to achieve 5w/1khz and run a frequency response Sweep, and then run multitone and so on. The data could be in CSV format or possibly saved as a png or jpg trace. I was unable to see any examples that showed how to do that- is it possible? Thanks for your time!
Hi @Var, when you enabled the CSV logging, all the tests will be logged to that file (and each test setup can have a different CSV name).
Inside the CSV, the data will be written with the following columns:
Note that you can see the pass/fail result, along with the measured result and the test limits that were in place when the measurement was taken. This is where you could also do some analysis on metrics such as gain and start to get a statistical understanding on where your limits should be.
The CSV logs are stored in a directory called “CsvLogs” off the
<mydocs>\QuantAsylum\Tractor directory. There’s also a
TestLogs directory off the
Tractor root. In there you’ll find a
index.html file and you’ll see every test that has ever been run, including pass/fail and the measured results. You can also click on the link to the screen shot and you’ll see the image captured for the measurement.
For the test you are looking to do, I’d think about it another way. Instead of trying to find the input level to hit 5W, look instead to measure the gain. And then take a few measurements at your gain limits. That should cover the range from just under 5W to just over 5W, and verify you are hitting your targets there.
It would be nice for a test to emit a value and allow that to be used for other tests (eg measure gain, and compute the output required to hit 5.00 watts), but that can’t currently be done.
Thanks for replying back and giving me the basics of how to do this. I do not have no limits for pass or fail, i just care about the data for what I am doing and I will probably do a manual measurement to get the say the 5w value and have a prompt to enter it at the start of the testing. Will I have to specify p/f limits? Sounds like it…
Hi @Var, are you aiming to test a bunch of different amps? Or a single amp over and over after you change something?
If the former, then I’d suggest you pick a sample of 10-20 amps to start characterizing. At that sample size, your margin-of-error is small enough that you can start making meaningful extrapolations from the, say, gain variations you saw across the 10-20 amps. And then from these measurements you can determine your pass/fail limits.
If the latter (eg tweak/measure/tweak/measure) then just set the pass fail limits really wide and run everything over and over.
For determining where you want things for manufacturing, the link below goes through some of the math associated with the problem.
Thanks for the feedback, Matt. What I am doing is testing a single amp, let’s say a power amp for this case, and I run similar tests on them- there are no criteria per say. I just thought it would be easier to try automating the procedure like I did when I worked using HP/Keysight/Agilent VEE. I test all kinds of amps- yesterday it was was vintage Jeff Rowland monoblocks and before that it was a vintage Crown amp. But I end up doing the same kind of tests on them so I thought it would be nice to trying to automate- plus I miss the programming. I have a small youtube channel- think a poor man’s ASR- where I look at how well audio gear has aged and other things that I think may be interesting.
Hi @Var, please share the link if you’d like. I think the interest in vintage stuff is very high, including tape decks.
Sure, I will post a link to one of my videos below. I am coming up on doing this for a year- so far I think I have posted one video a week, and for most of them I am using my QA402 and do mention that from time to time. I have learned a lot from doing the measurements, but am always open to suggestions/ criticisms and looking back realize that I made some errors with some of my data taking. I have several local audio friends (hoarders ) who are willing to let me test and repair their gear as needed. Most of what I have looked at is solid state- this video was the 1st tube gear I had looked at. It is a great outlet for my “audio hobby” and helps keep me exercising my “grey cells” as Hercule Poirot would say.
Vintage Audio Review Episode #43: McIntosh MC240 Power Amplifier - YouTube
Hi @Var, man, for an amp made in '61 it’s awesome with a bit of TLC it’s back to spec. What is the highest DC voltage you’d encounter under the hood? I think your impedance measurement is correct, because the switch from 8 to 4 is looking only for the voltage change. From that the output Z is learned.
If you wanted to know the output Z when using the 4 ohm taps, then you’d do the same: Connect to the 4 ohm taps, let the HW switch from 4 to 8 ohms to see how amplitude changes, and then that would tell you 4 ohm output Z.
Do the transformers ever go bad in old hardware if they haven’t burned up due to mistreatment? In other words, if you treated your equipment properly, the caps will probably need TLC after decades. But would the transformer?
PS. Love the Realistic AP500 video! I still have Radio Shack catalogs from the 70’s. My first receiver was an STA-52B. I wish someone that was there would do a book about what the engineering teams were like at Tandy in Forth Worth in the 70’s.
Good to here about the impedance measurement being ok. I still measure much less than what most amps were spec’d at damping factor wise, which is 8/impedance. The last amp I looked at was 80 and the spec was like 500- 80 is pretty darn good, though. That was my only tube amp that I looked at, though I have a smaller 25w McIntosh on the bench now that I am trying troubleshoot- one channel starts distorting too much too early and I have had very little experience with tubes. I use a x10 scope probe hooked up to one input of the QA402 for troubleshooting at times- but not with more than about 125vdc (at least today). Sometimes I get good results, but not so much with the tube stuff. Most I have measured (with a scope or dmm) was about 500vdc, but 400vdc is more typical. I have not run into transformer problems so far, but may have a power supply transformer going bad in a tube integrated amp I worked on and set aside to look at the McIntosh. I am not one to to replace the caps just because something is “old”. THe STA-52B made it to the Classic Receiver Website! It would be interesting to know how their engineering group operated, though I think they had other companies (not popular ones like Pioneer and Kenwood) make stuff for them. BTW, I posted a review on the ASR website the other day, of an A/B switcher I purchased, and describe using the QA402 to test it:
Hi @Var, an experiment you can do to double check is as follows:
- Drive your amp into a load R1. Measure the amplitude across R1. Call this A1.
- Drive your amp into a load R2 with the same level as 1). Measure the amplitude across R2. Call this A2.
The output impedance can then be calculated:
Z = (R2 * R1 * (A2 - A1)) / (A1 * R2 - A2 * R1)
The derivation is as follows:
A1 = Vout * R1 / (R1+Z)
A2 = Vout * R2 / (R2+Z)
Then, you can put these two equations into Wolfram Alpha and ask it to solve for Z:
It will give the answer (among others):
So, using two arbitrary resistors and two amplitude measurements, you can learn the output Z of an amp if you ever need to double check
Thanks for taking the time to explain this more- have not had to do algebra for a while, but this is not too bad I am also wondering if the method/equipment that was used 30-40years ago to measure output Z may not have been as accurate at what the QA40x is doing… BTW, the QA402x helped me find a pair of bad output tubes in an amp that I was working on- pretty cool and will be in an upcoming video.
Is there a location on the website, or here on the forum that has all of the application notes like this one collected into one location? Location?
Hi @Var, the method for measuring output impedance has been the same for a long time: Apply an arbitrary load and see how the output level changes. The equations above simplify a lot if your R1 is infinite. And then, you can also use a expo sweep and compute the impedance over frequency instead of at a single tone.
On the opamp tester HERE, note there are three load resistors (4 if you include open). From those the output Z of the opamp can be easily learned, and for closed loop gains they aren’t usually published. But they can approach 1 ohm in some configurations. And this all ties together as you aim for low-noise opamp circuits (with low Rs) but not so low that the driving opamp impedance starts to matter.
Hi @CESAUDIOPRO, unfortunately not. I think the more detailed topics that require further discussion are put over the in Test Plugins Wiki on Github.
And the more general Wiki related to the QA40x is located
The aim is to move the functional details to the manual (for example, “this button does this”) and then move the how and why the github locations above. No question a lot more could be done.
Thanks the added info. I thought about what you posted earlier and what I have seen is that the damping is specified for one frequency, typically low- for one amp it was 20Hz. Next time I measure one it will do it my normal way with the QA402 and then put in the specific specified frequency and measure it using my scope and see how it compares…
Update- I tried using the 4ohm tap instead of the 8ohm tap on a McIntosh tube amp that I was testing- hooking up an 8ohm load to the the 8ohm tap and then hooking the 4ohm load to the 4ohm tap for the measurement, and got about 9ohms for the output impedance. When I hooked up the 4ohm load to the 8ohm tap, as I had done previously, I got about 0.8ohms. I did not try a manual measurement (yet)…
Hi @Var, remember the equation only works if you are only changing one variable (the load). If you change more than one thing, then you have too many unknowns for the result to make sense.
A good experiment to try might be setting up to the 8 ohm tap, and then measuring at 8 ohms load and 8.1 ohms load (or better still the 7.9 and 8.1 ohms). That would keep the output transformer happy because it’s seeing effectively the same load. The QA451 amp load plug-in can only do 4 and 8 ohms, but if you use the PwrOutputImpedance plug-in then you can specify the loads.
And then repeat for the 4 ohm tap with 3.9 and 4.1 ohms.
I will see if I can put together some more precise loads and see what happens- thanks for the suggestion!
Impedance measurement update- I finally was able to check the output impedance measured by the QA402’s PwrOutputImpdeance test vs one made manually using the equation you listed earlier. I only measured one channel and only at 1khz, and measured the voltage going to the load with my new DMM. I created a little spreadsheet and adjusted the resistor values slightly for the speadsheet vs. what I had in the PwrOutputImpdeance test options. Below is how the QA402 was set up for the measurement (-7.9dBv yields 5w/8ohms):
Here is the plot:
For the right channel at 1khz the output impedance is 0.13ohms. My manual measurement using 8.17ohms & 4.11ohms for the resistor values was .133ohms. Pretty good agreement. This translates to a damping factor of 60, way less than the >200 spec which has always been the case with any damping factor I have measured… But, there may be a problem somewhere with the low frequency measurement done by the QA402, as I measured it manually at 20Hz, and it was about .13ohms, and not the value that seems to be much greater than that as shown by the graph. I tried the measurement a few times and the 20Hz value not correct- any ideas on what may be causing this problem there…?
Hi @Var, see if you can run the amp output impedance test and try to sweep from, say, 5 Hz to 25, with a 5 Hz step. And see if that improves things at 25 Hz. Sometimes, you can be dealing with a smart amp that will take a beat to turn off noise gating (sounds like your amp doesn’t have that). But you can also be dealing with DC offsets working their way through the system. Also, if your FFT is too small, you might not catch enough full cycles.
After trying a range of freqs at the low end, then try just 20 Hz by itself and when it finishes the single test, take a look at the time domain. That might give a clue where things are falling apart. You can also just run manually at 20 Hz, and verify the amplitude readings are making sense (FFT size, windowing, etc)
Thanks for the reply back. The amp is not that smart- a 1990 McIntosh. There are a few other tests that I have to do over a wider band than 20hz-20khz- the impedance test normally works fine. The FFT size is probably 128k. I will try it again and let you know what I find out.