DIY THD Tube Tester / Analyzer - diyAudio
Go Back   Home > Forums > Amplifiers > Tubes / Valves

Tubes / Valves All about our sweet vacuum tubes :) Threads about Musical Instrument Amps of all kinds should be in the Instruments & Amps forum

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 25th October 2014, 03:03 AM   #1
diyAudio Member
 
Join Date: Oct 2014
Default DIY THD Tube Tester / Analyzer

Making a DIY THD vacuum tube analyzer with a Beaglebone or Arduino is of interest and although I've done small electronic projects this is clearly more ambitious for me; but I'm determined to make this and so I joined today hoping that I could get some kind advice from you.

I did a ton of searching for a THD chip and the MSTHDA chip from Mixed Signal Integration is the only THD Analyzer Chip that I could find and so I have them now to work with. Here is the link for the MSTHDA Data Sheet.

If you know of some other chips to consider please let me know; for now I'll focus my questions to the MSTHDA chip.

My thought is to run a sine wave signal into the tube and gradually increase the voltage seeking to find where the THD starts dramatically increasing; I'd gradually increased the frequency after each test, repeating until I'd scanned through all frequencies ranging from 10Hz to 25KHz.

I would appreciate suggestions how to accurately calculate the THD output from the tube and coaching for selecting the best ADC chip. Should I sum all 5 harmonics together and compare them to the original MUX signal? Should I compare the primary harmonic with the sum of the other 4 harmonics?

Thanks for your help, as you can see I need it.

Mark
  Reply With Quote
Old 25th October 2014, 04:16 AM   #2
Keit is offline Keit  Australia
diyAudio Member
 
Join Date: Feb 2014
Location: Australia
I take it from your first sentence that you want to build a tester that measured the distortion of a tube, ie a tube tester, and not testing ditortion in a complete amplifier circuit.

You are actually going about an easy task a hard way. That's because the anode current in a vacuum tube is related fairly accurately to grid voltage by a simple mathematic law - the so called three-halves-power-law, up to the point of cutoff or saturation.

The impact of this is that triodes on their own produce only second harmonic distortion to any significant degree. The fraction of 2nd harmonic compared to the fundamental is zero for zero signal level. rising smoothly with signal level.
Pentodes are only slightly more complex in this regard.

Higher order even harmonics and odd harmonics only occur, for all practical purposes, in-circuit after negative feedback is applied.


Checking distortion in a complete amplifier is another matter. Additional distortion is cuase in transformers, stray feedback in power rails, etc. Negative feedback lowers overall distortion, but generates additional higher order harmonics. Errors in setting bias is shown up by both power output tests and distortion tests.


Note that for historical reasons connected with hif-fi magazines and reviewers, the average man in the street and amateur constructors think its all about harmonic distortion. But what the ear is realy sentive to is intermodulation distortion. Intermod distortion is what profession engineers always checked, and is how professional electronics for recording studios, radio stations, and motion picture equipment is specified.


The amount of distortion introduced by a vacuum tube is quite independent with frequency, up to the point where the time taken for the electrons to travel from cathode to anode becomes significant compared to the time of one cycle. Since this is nanoseconds, only engineers concerned with frequencies above the high megahertz regoin need worry about it.

All the above is why tube testers capable of testing tube distortion were never sold during the pre-transistor era. Datasheets give the power output for (typcally) 5% distortion - the onset of clipping - that's all you need to know. The distortion arising within a tube at any given non-overload signal lvel is trivially easy to calculte, and that's what tube amp design engineers did.

But if you DO what to make a tube distortion tester, and doing measurments is a great way to bed in your theorectical knowlege, all the above means there is a simple way: Inject a clean sinewave into the grid at a convenient frewquency - say 5kHz. Measure the anode current swing by measn of a curent-to-voltage op-amp connection to get output signal amplitude. Use a tuned circuit of filter to extract the second harmonic (10 kHz - measn another op-amp or 2) and measure that.

What commercial tube testers (the better ones) did is measure power output for power tubes. There's little or no point in establishing the precise point at which overdriving causes distortion to rise above a certain point, as it isn't precise. It varies from tube to tube, and falls off with use as the tube cathode looses emission over time.

Last edited by Keit; 25th October 2014 at 04:28 AM.
  Reply With Quote
Old 25th October 2014, 07:22 AM   #3
Merlinb is online now Merlinb  United Kingdom
diyAudio Member
 
Merlinb's Avatar
 
Join Date: May 2006
Location: Lancashire
Mark, if you go looking for a sharp rise in distortion you may be disappointed! THD in a simple non-feedback tube circuit often rises very smoothly all the way up to, and beyond, clipping. You may simply have to measure THD all the way up to 10%, say, (40% is basically pure square wave) before retracing.


Quote:
Originally Posted by Keit View Post
Note that for historical reasons connected with hif-fi magazines and reviewers, the average man in the street and amateur constructors think its all about harmonic distortion. But what the ear is realy sentive to is intermodulation distortion. Intermod distortion is what profession engineers always checked, and is how professional electronics for recording studios, radio stations, and motion picture equipment is specified.
This is not true. Although it can be argued that IM is a better test of distortion audibility, it is almost never tested, even on studio equipment (radio is different, however). This is because IM is proportional to THD, so as long as you know one, you have a pretty good idea of the other, except THD is a LOT easier to measure. There are many different possible tests for IMD, so few companies will bother with it unless there was firm universal agreement. THD v level and THD v freq' is quite sufficient for professional work. In fact it is usually the audio magazines and reviewers who shout for IM distortion measurements, not understanding the intricacies of the problem!

Last edited by Merlinb; 25th October 2014 at 07:31 AM.
  Reply With Quote
Old 25th October 2014, 12:28 PM   #4
Keit is offline Keit  Australia
diyAudio Member
 
Join Date: Feb 2014
Location: Australia
Quote:
Originally Posted by Merlinb View Post
This is not true. Although it can be argued that IM is a better test of distortion audibility, it is almost never tested, even on studio equipment (radio is different, however).
This is because IM is proportional to THD, so as long as you know one, you have a pretty good idea of the other, except THD is a LOT easier to measure.
You seem quite confused.

1. Yes, on studio equipment it is (nearly) always tested. I've worked in the industry - I've tested it lots of times. Its what we do.

2. IM level is closely related to THD level only in simple active devices having no frequency dependent phenomena and no feedback, local or overall. Real amplifiers have a myriad of frequency dependent phenomena - interstage coupling, shunt capacity, transit time effects in semiconductors. And there are a variety of feedback mechanisms, not just the one the designer put in! For example, consider an amp just on the point of suffering from slew-rate limiting. The THD may be fractional percentage, but the IM can be gross - tens of percent - and the amp will sound gross too!

3. Actually IM is easier to measure. To measure THD you need a precision low distortion oscillator, and the design of the measuring set must also be very good. In fact, for accuracy, the THD of the oscillator and measuring circuit musbe better than 10x lower than the amp you are testing. To test IM you need two oscillators, but neither needs to do anything better that roughly approximate a sine wave. You need some other circuitry, but none of it needs any precision.

Hum and noise confuses and confounds THD testing, but matters not a whit in IMD testing.

4. There are a variety of ways to measure IM, and they all give different results to geater or lesser degree (which is a proof the IM is not proprtional to THD), however the metod of measuring it was standardized by the professional sound reproduction industry a long time ago. See any edition of the SMPTE Manual (SMPTE = Society for Motion Picture and Television Engineers) for example. SMPTE is American. Standards bodies in other countries have also set out how to do it. For example, I have tested radio station studio/transmitter links by the European CCITT method.

When a professional or capable amateur engineer is testing the design of a new amplifier, THD is a valuable test, as by looking at the distortion residual, it is very easy to diagnose what's wrong with the amplifier. For example if the distortion rises at or near zero crossing, you have problem with crossover distortion. So you should look at bias. If the distortion is greater positive peaks than it is on negative peaks, say, you may have a problem with Early effect. No good playing with bias then! THD is particularly valuable in diagnosing design faults in solid state amplifiers. It is less usefull in diagnosing design faults in tube amps, but still good to do.

Unfortunately IM is not much use in diagnosis. It will tell you an amp is bad, but not why it's bad, like THD can.

But diagnosing design faults has nothing to do with how an amp sounds to the end user. THD is a poor indicator of how good an amp sounds. IM is very good.

It should be noted that when technicians are repairing an item of audio equipment, ie with component faulst and not design faults, they almost never test THD or IMD. They don't need to. If a circuit design is sound, it will do what it does, well, or do it so badly more simpler tests will show it up. For example, for a tube amp, if the power output is normal, there cannot be much wrong with it.

Last edited by Keit; 25th October 2014 at 12:53 PM.
  Reply With Quote
Old 25th October 2014, 01:53 PM   #5
DF96 is offline DF96  England
diyAudio Member
 
Join Date: May 2007
There is a fundamental flaw in a 'tube THD' analyser: a tube does not have a THD; a tube with a signal does not have a THD, either. To get a THD you need three things: a tube, a circuit, and a signal.
  Reply With Quote
Old 25th October 2014, 03:43 PM   #6
diyAudio Member
 
tubelab.com's Avatar
 
Join Date: Jul 2005
Location: West Virginia panhandle
A tube may not have THD, but a tube can be noisy, and noise is distortion, but not harmonically related. The THD filter chip here will not catch a noisy tube.

Most THD analyzers built before computers, fed a clean sine wave to the circuit under test, then ran the circuit's output through a notch filter to remove the original sine wave. Everything left was measured and called "THD." This type of measurement is now referred to as SINAD (SIgnal to Noise And Distortion) and is probably what you want here.

I started sorting through about 100,000 surplus tubes about 10 years ago. I started down this road with a pair of well regarded mutual conductance tube testers. Within a few years I discovered that tubes that tested good in both testers could be noisy, or exhibit high distortion in a working amplifier. In fact the tube testers proved rather useless on large signal audio output tubes since they did not test the tubes in real world operating conditions. I sold both tube testers.

After multiple iterations of DIY testers, I settled on a simple setup that worked well. I never built a "tube tester" but just set the system up on my breadboard as needed, since I would typically "test" several hundred of the same type number at a time.

For small signal types I breadboarded a simple RC coupled amplifier fed with a clean sine wave audio oscillator (HP204C). I used a variable power supply with a bias output that controls the tube's grid. The output of the test circuit is fed to a scope, a THD analyzer (HP8903), a sound card based FFT analyzer, and the input of a guitar amp (or other amp with a speaker). The power supply for the stage under test has meters for current and voltage.

Upon plugging in a fresh sample to be tested, leave the audio oscillator off, and the guitar volume up. Listen to the tube's output during warm up for strange noises and hiss. Watch the current meter on the power supply, it should rise steadily and stabilize within 30 seconds or so. Look for "creepers" that slowly keep climbing. They are likely gassy. Then whack the tube with a pencil and listen for microphonics. Then switch off the amp, and turn on the oscillator. Check the THD, and look at the spectrum display on the FFT.

After testing a few (OK a few thousand) tubes in this manner you know what to look for and where to set the oscillator (just below clipping on an average good tube), and how much hiss a particular tube should make. You test under different voltage and current settings for tubes that will see a wide variety of operating conditions. An amplifier tube will work form DC well up into the MHz range so testing at different audio frequencies is not necessary.

For large signal tubes that will be used for audio I set up a simple SE amplifier with an OPT and performed the same testing as with small signal tubes. I typically turn the bias voltage down to raise the tube current while watching the distortion. I test up to and beyond the rated cathode current to catch worn out cathodes that a tube tester will not find.

For rectifier tubes I run DC current through each plate to cathode individually and measure the forward voltage drop. Compare to the published curves.

I got boxes full of loose untested used tubes that had been saved since WWII. They were pulled from military spares, and scrap. In some cases there were thousands of each type. After having mixed results with a tube tester, I started using this method. I found that as many as 50% of the tubes tested in this manner were actually bad. The average was about 20% bad.

I got another lot of NOS tubes from the 60's that came from a defunct TV parts warehouse. The same type of testing found from 5 to 20% bad tubes, depending type, and how they were stored....some were broken.

You may not need or want to get this involved, or maybe you can figure out how to automate this process....I have already gone through most of my tubes, and given away all the tubes I didn't want to test.
__________________
Too much power is almost enough! Turn it up till it explodes - then back up just a little.
  Reply With Quote
Old 25th October 2014, 04:08 PM   #7
Keit is offline Keit  Australia
diyAudio Member
 
Join Date: Feb 2014
Location: Australia
Quote:
Originally Posted by DF96 View Post
There is a fundamental flaw in a 'tube THD' analyser: a tube does not have a THD; a tube with a signal does not have a THD, either. To get a THD you need three things: a tube, a circuit, and a signal.

The tube tester supplies the circuit. Further, the relationship between anode current and grid voltage is in accordance with a power law. Hence harmonic distortion is a property of vacuum tubes. In circuit, the distortion may be less or it may be greater than the tube's inherent distortion.

Your statement is just like saying a tube has no amplication unless in a circuit. But tube manufacturers have always quoted circiuit-independent tube parameters that express the tubes cability to amplifer - the gm (ratio of anode current swing to grid voltage swing) and mu (ratio of anode voltage swing to grid swing). A tube's THD is a measure of the variation in gm with instantaneous signal amplitude, and is a perfictly vaild and useful thing to know.

Last edited by Keit; 25th October 2014 at 04:14 PM.
  Reply With Quote
Old 25th October 2014, 08:28 PM   #8
diyAudio Member
 
Join Date: Oct 2014
Quote:
Originally Posted by Keit View Post
I take it from your first sentence that you want to build a tester that measured the distortion of a tube, ie a tube tester, and not testing distortion in a complete amplifier circuit.

You are actually going about an easy task a hard way. That's because the anode current in a vacuum tube is related fairly accurately to grid voltage by a simple mathematical law - the so called three-halves-power-law, up to the point of cutoff or saturation.

The impact of this is that triodes on their own produce only second harmonic distortion to any significant degree. The fraction of 2nd harmonic compared to the fundamental is zero for zero signal level. rising smoothly with signal level.
Pentodes are only slightly more complex in this regard.

Higher order even harmonics and odd harmonics only occur, for all practical purposes, in-circuit after negative feedback is applied.


Checking distortion in a complete amplifier is another matter. Additional distortion is cause in transformers, stray feedback in power rails, etc. Negative feedback lowers overall distortion, but generates additional higher order harmonics. Errors in setting bias is shown up by both power output tests and distortion tests.


Note that for historical reasons connected with hif-fi magazines and reviewers, the average man in the street and amateur constructors think its all about harmonic distortion. But what the ear is really sensitive to is intermodulation distortion. Intermod distortion is what profession engineers always checked, and is how professional electronics for recording studios, radio stations, and motion picture equipment is specified.


The amount of distortion introduced by a vacuum tube is quite independent with frequency, up to the point where the time taken for the electrons to travel from cathode to anode becomes significant compared to the time of one cycle. Since this is nanoseconds, only engineers concerned with frequencies above the high megahertz region need worry about it.

All the above is why tube testers capable of testing tube distortion were never sold during the pre-transistor era. Datasheets give the power output for (typically) 5% distortion - the onset of clipping - that's all you need to know. The distortion arising within a tube at any given non-overload signal level is trivially easy to calculate, and that's what tube amp design engineers did.

But if you DO what to make a tube distortion tester, and doing measurements is a great way to bed in your theoretical knowledge, all the above means there is a simple way: Inject a clean sinewave into the grid at a convenient frequency - say 5kHz. Measure the anode current swing by means of a current-to-voltage op-amp connection to get output signal amplitude. Use a tuned circuit of filter to extract the second harmonic (10 kHz - means another op-amp or 2) and measure that.

What commercial tube testers (the better ones) did is measure power output for power tubes. There's little or no point in establishing the precise point at which overdriving causes distortion to rise above a certain point, as it isn't precise. It varies from tube to tube, and falls off with use as the tube cathode looses emission over time.
I got the idea while watching a few videos on YouTube done by Elpaso TubeAmps. Here he shows his equipment setup and here is a KT88 test, one of several tests that he does. You'll notice that he increases the voltage until he see's the sinewave start to break up and then measures the THD. What I also noticed in his test is that you reach a certain voltage where a "knee" type pattern occurs with the THD, where there is a dramatic increase in distortion, it just starts to skyrocket out of control compared to a gentle linear looking rise.

I felt this could be duplicated by plotting out the THD to see where it starts to break away from a gradual rise to a fast rise.

Notice in his test he is using a pair, however his probe is plugged into a tube socket on the side which made me wonder if he's measuring prior to any amplification that would occur with a second tube; I'm not familiar with the Dynaco Mark III, someone else here is I'm sure.

I have 2 goals with this, mainly I want to learn more about THD (and THD+N, it's still confusing to me), and I really like the idea of evaluating a tube with test measurements as best that we can in a controlled, repeatable, dependable manner. I appreciate any suggestions on how to best do some test measurements on power tubes especially, because I'd like to do this using a Beaglebone or Arduino so that I can capture data and plot things out to look at them and learn.

Thanks again for the suggestions.

Mark
  Reply With Quote
Old 25th October 2014, 09:26 PM   #9
diyAudio Member
 
jazbo8's Avatar
 
Join Date: Jan 2011
Location: In Transient
Read Patrick Turner's article and see if you can integrate uP control into the test rig.
  Reply With Quote
Old 26th October 2014, 01:00 AM   #10
Keit is offline Keit  Australia
diyAudio Member
 
Join Date: Feb 2014
Location: Australia
Quote:
Originally Posted by markmain View Post
What I also noticed in his test is that you reach a certain voltage where a "knee" type pattern occurs with the THD, where there is a dramatic increase in distortion, it just starts to skyrocket out of control compared to a gentle linear looking rise.
Yes, that's the point where the tube is overloaded - the point where the grid is cutoff and/or where the anode is bottomed - the anode voltage cannot go much lower. So the output can no longer follow the input - that's the definition of distortion. Nothing remarkable about it. The overload point can be determined much more simply by measuring the power ouput. That's not a precise as indentifying at what signal level the distortion rises to a stated percentage. But as I said, precision is pointless, as the overload point changes over the life of the tube. And a power output increase of 10% is barely audible anyway.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
LTSPICE THD Analyzer Eugene Dvoskin Software Tools 24 11th October 2014 07:53 PM
DIY tube tester Joe Lebowski Tubes / Valves 8 31st March 2013 06:52 PM
Chip THD% Analyzer jackinnj Equipment & Tools 1 27th December 2007 10:28 PM
thd analyzer-on-a-chip unclejed613 Everything Else 0 2nd January 2007 11:18 PM


New To Site? Need Help?

All times are GMT. The time now is 03:14 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2