advertisement


Analysing your diy builds

B

brickj

pfm Member
Having built a few pre amps, headphone amps & power amps, I have the urge to measure their performance more quantitatively - even going as far as looking at HP 8903a/b audio analysers (why are they 2x in the UK!?).

Has anyone measured the performance (THD, frequency response, power output etc) of their StarFish, B4, HackerNap, latest-diy-project?

I've experimented using my Mac's soundcard for some basic analysis, but I was never certain of the calibration and whether I was seeing artifacts from the soundcard vs. the device. Also, at the time, the more readily available PC software wasn't happy running on a Virtual PC. I would much prefer a piece of old-school test equipment.

Curious to hear what others have found to be successful and any results they would care to share.

Thanks.

Justin.
 
I have used software called Baudline, running under Linux on a laptop.

For some measurements I use my HP DVM, as it has a true RMS AC capability.
 
i have a blackstar low distortion oscillator and a picoscope 216 for doing distortion and noise measurements.

I have a neutrik rt-1m gpib test rig that I am still working out how to use.

I also have ARTA real time analysis software that in conjunction with my delta 1010 sound card can produce some nice sweeps.

I have recently been investigating 'speaker workshop' that will make LCR and impedance curve measurements with your soundcard and a reference resistor.

ultimately the most flexible test equipment you have are your own ears.
 
I tend to look at things in an oblique way. I'm generally not too interested in hitting a %THD or S:N benchmark, but understanding what is going on - the character of that noise or distortion and so forth.

So mostly - a good analogue 'scope, a soundcard and bits of free software for spectrum analysis - really, nothing fancy (also have / used Spectralab suite).

Most useful is knocking-up a simple DIY measurement preamps of 20dB (10x) and 40dB (100x) gain, sometimes more. It make other tools really useful, by lifting the target signal into range of good resolution from even cheap instruments or software. Some judicious input filtering of the signal is very helpful to limit measurement bandwidth to match what is being hunted. Sometimes that means tacking-in a tiny input coupling cap to only look at noise/harmonics above audio; sometimes it means restricting input bandwidth to 10-20Khz to look only at intermod within the audio range and so on.

There are loads of simple 3-opamp 'inamp' schematics out there, I'd recommend it as a DIY build (perhaps worth a thread here, actually...). But even the simplest approach can be a huge help: take an NE5532, a couple of 9volt PP3 batteries (to avoid ground loops), build it 'dead-bug' style on a piece of blank copperclad pcb with suitable sockets or shielded flying leads and configure it for gain of say 10x with the input bandwidth limited to say 20-30Khz (RC on the input) and bingo - your scope or soundcard is now good for signals down to 100uV or even better. You start noticing you have to get very crafty about shielding and noisy things nearby, like the PC or smps-powered items. Use a small coil of wire or loop (eg bent paperclip) on the end of a piece of coax instead of a 'scope probe and you can even qualify magnetic field emissions too...

NB with a similar approach, a biscuit tin, a sound card and no little cunning Andy Weekes did noise measurements on the development of the PFM Flea here that had a noise floor down below -160dB relative to 1v to beyond 30Khz. That's just 5nV!
 
The other side of course is suitable test signal sources.

I use a simple bit of sig gen software on my ancient (kitchen) iMac to generate sines, noise, squarewaves and other arbitrary signals. Broadband white noise when wanted is from the classic source of a reverse-biased small signal transistor base-emitter junction run at very low current (I use a BC557, and a regulated12v supply through 560Kohms; the junction 'zeners' at around 6-7v ish and is nicely noisy - 10x opamp buffer brings it up to useful levels)

Eitheer way, a simple opamp based buffer with volume pot and a dab of (adjustable with soldering) bandwidth-limiting does a 'good-enough' job of cleaning -up the signal/setting level/ desired source impedance and so on, as required.
 
If you are interested in super low noise testing look at Cyril batemans series of articles on capacitor sound.
 
I built two of Cyril's test sets and the oscillator output certainly is very pure. But a single frequency is a bit limiting. The rig also makes a good vibration/piezo sensitivity tester. Re distortion I found much the same as Cyril. Around that time Doug Self in EW published a plot of distortion vs frequency for an Electrolytic cap using an Audio Precision test set. It had a big hump in the middle of an otherwise gently upward sloping trace. So it became clear that for some caps at least, spot tests at one or two frequencies wasn't enough to characterise a cap.

At that time on pfm, tants seemed to be able to hold their own even though having relatively a lot of distortion. My guess was that subjective cap differences were due to the way ESR varied with frequency.

Rather than a star earth scheme Cyril uses a wide track around the outside of the pcb and runs the 0V connection of every supply decoupling cap and device input directly to the nearest point on it. Given the stunningly low distortion and noise, there is obviously merit in the approach.

Cyril was into his 70s back in 2004, and disgruntled at how little Electronics World were willing to pay for his Capsound series. I'm not surprised he hasn't been heard from since.

I have a Fluke function generator from the '80s I bought at a Defence surplus auction. It is very flexible and the sync. signal that you connect to a scope's external timebase input is great when your signal is buried in noise etc. It has a 40mV DC offset, which is present even when DC is nominally off, the discovery of which wasted several hours. This is actually within spec.
 
Cyrils use of dougs 'blameless' amplifier in testing speaker cables amused me when they blew up.

Well it wasn't the amps fault!
 
David you tease. Even today I have all those EW Capsound copies, 10 by my count, around 3m from my desk. I vaguely remember him testing caps in amps.
(shuffle, shuffle)
A quick search revealed only this:
"Using the workhorse Maplin Mosfet 100W mentioned above...."
in his article in EW October 2003.

P.S. I'd be interested to hear what you're doing with GPIB. I bought an Agilent GPIB-USB adapter a while back to aid data capture from my Wayne-Kerr Component Analyser. I didn't get past command line use and I've had a bit of a DIY hiatus over the last 18 months.
 
i have an old pc with an ISA slot to accomodate my NI GPIB adaptor which is then connected up to the neutrik rt-1m.

neutrik kind of forgot all about the unit but over on diyaudio a few guys have been working on it.

the article on exploding amplifiers is this one:

http://www.waynekirkwood.com/Images/pdf/Cyril_Bateman/Bateman_Speaker_Amp_Interaction.pdf

and it amazes me how much importance there is at RF frequencies with respect to audio amplifier stability.


PS: very impressed that you made up the capsound test gear. would love to see it!
 
Dave,
Many thanks. I'll settle down for an enjoyable read in the next day or two.

I'll post a pic of my capsound rig after I've reactivated my Flikr account. I'm not set up to actually turn it on at the mo. Analog released an upgraded voltage controlled amp IC which I bought and should try.

Where abouts are you? Unless you're an Aussie like me it must be dark outside.

We have talked before on GPIB:
http://www.pinkfishmedia.net/forum/showthread.php?t=17765&highlight=neutrik
 
well remembered!

its late spring here so days are getting longer and the rain that comes with wimbledon beckons.

I'll take some photos of the gpib rig (that i didnt dispose of) and see if i can get some steam up in the old pcs innards.
 
Thanks for the information. Unfortunately I've not had much time this week, but the baudline software looks just what I was after and is clearly well designed and has a Mac OS X port, which is a bonus for me. I've briefly played with it using this sound card interface, which I built sometime ago. This will help with my urge to measure a few characteristics of my builds!

I also hadn't come across those cap-sound articles before, they're a great read - although I can only claim to have digested the first couple of articles so far. I personally appreciate the thoughtful and scientific analysis.

I'll post some pics from baudline as soon as I have some.
 
You must log in or register to reply here.


advertisement


Back
Top