Luke's basic amplifier tests

[Ricci]

Typically an amp will have a voltage limit set by the rails and power supply. Say you have amplifier X and it is capable of 50 volts prior to clipping. Since the amplifier operates as a constant voltage device the current demands increase into lower impedances and at some point the amplifier cannot provide enough current to meet the demand to maintain the voltage required at say something like 1ohm. The current is the limitation. If the impedance is higher say 16ohms or something, the current demand is a small fraction of what it was into the lower impedance and the amp is not being asked for much power. In that case you will be limited by the voltage capability of the amp. There is a sweet spot of overlap where most amplifier are able to produce maximum voltage, current and power. It is odd to see a lower voltage limit at a higher impedance. I can't recall seeing that in any amplifier test before. The voltage limit should be constant to my knowledge once it is the limitations instead of the current.

 

I am not an expert on amplifiers but this is my understanding of modern amplifiers.

[lukeamdman]

Typically an amp will have a voltage limit set by the rails and power supply. Say you have amplifier X and it is capable of 50 volts prior to clipping. Since the amplifier operates as a constant voltage device the current demands increase into lower impedances and at some point the amplifier cannot provide enough current to meet the demand to maintain the voltage required at say something like 1ohm. The current is the limitation. If the impedance is higher say 16ohms or something, the current demand is a small fraction of what it was into the lower impedance and the amp is not being asked for much power. In that case you will be limited by the voltage capability of the amp. There is a sweet spot of overlap where most amplifier are able to produce maximum voltage, current and power. It is odd to see a lower voltage limit at a higher impedance. I can't recall seeing that in any amplifier test before. The voltage limit should be constant to my knowledge once it is the limitations instead of the current.

 

I am not an expert on amplifiers but this is my understanding of modern amplifiers.

 

I completely agree on all counts, but I think with the XLS and SP2 I was hitting the voltage limits.

 

Without any load connected I can only get ~100v out of the XLS, and the spec sheet for the SP2 says ~126v from what I recall, which would explain the clipping I saw on the scope. 

[SpeakerPower]

Ricci is right. And rails typically will sag as current ramps up. I've never seen higher maximum voltage at lower load impedances either.

 

brian

[SME]

My suspicion lies in the measurement device, which in this case is the scope.  In theory, the scope should be able to measure True RMS down to ULF very accurately, being that it does digital sampling.  The trouble is that the actual algorithm implemented on the scope may not do this well.  I honestly have no idea if this is really the case, but it seems plausible.

 

Can the scope report the peaks instead?  If the waveform is undistorted and you are always testing with sine waves, then you can back calculate Vrms from Vpeak by multiplying by 0.707.

 

Edit: I mean multiplying, not dividing.  (*head smack*)

Edited March 19, 2016 by SME

[lukeamdman]

Kind of an impulse buy, but I got a used Fluke 87V off eBay.  It has a frequency range of 0.5hz-200khz, so we'll see if it oscillates below 20hz like the 117/115. 

[Ricci]

That's a nice piece of equipment.

 

Going back to the voltage results a bit. If it were me the first thing I would do is get a baseline for what you are using to record the voltages and remove variables. Forget about high power and voltage and using the speakers hooked up at first. Use a small signal like 3 volts straight out of the amp or a generator and check the readings at various frequencies. Baseline a 55Hz then try 100, 30, 20, 10, 5Hz etc. Any reasonable amplifier or generator should be flat 200 to 20Hz within a very small window. If your meter is measuring variation from 200Hz to 20Hz it is likely not accurate. 

 

If everything checks out ok there hook the subs and rest of the chain back up and try the same deal once again at a low level consistent voltage. Baseline at 50-60Hz again and change nothing other than the frequency being used. Make sure that you aren't getting weird voltage readings that don't make sense. It should be consistent regardless of frequency until well below 20Hz where you'll start to see a reduction at the lowest frequencies.

 

Also it looks like your impedance measurements might be done by hand and then plotted or something? Forgive me if that's not the case but I noticed that they look a bit odd too. Especially the Othorn one. It's missing a peak which only shows up as a tiny bump.

[lukeamdman]

That's a nice piece of equipment.

 

Going back to the voltage results a bit. If it were me the first thing I would do is get a baseline for what you are using to record the voltages and remove variables. Forget about high power and voltage and using the speakers hooked up at first. Use a small signal like 3 volts straight out of the amp or a generator and check the readings at various frequencies. Baseline a 55Hz then try 100, 30, 20, 10, 5Hz etc. Any reasonable amplifier or generator should be flat 200 to 20Hz within a very small window. If your meter is measuring variation from 200Hz to 20Hz it is likely not accurate. 

 

If everything checks out ok there hook the subs and rest of the chain back up and try the same deal once again at a low level consistent voltage. Baseline at 50-60Hz again and change nothing other than the frequency being used. Make sure that you aren't getting weird voltage readings that don't make sense. It should be consistent regardless of frequency until well below 20Hz where you'll start to see a reduction at the lowest frequencies.

 

Also it looks like your impedance measurements might be done by hand and then plotted or something? Forgive me if that's not the case but I noticed that they look a bit odd too. Especially the Othorn one. It's missing a peak which only shows up as a tiny bump.

 

I think this post got lost in the mix a few pages back:

 

http://data-bass.ipbhost.com/index.php?/topic/476-lukes-basic-amplifier-tests/?p=12560

 

I did run voltage tests without any subs connected with both the scope and the Fluke plugged right into the amp to compare the results:

 

 

 

As you can see, from 20-120hz the voltage only changes by 0.02v.  At 10hz and 5hz you begin to see my signal chain roll-off (Marantz AV7005 + MiniDSP 10x10) which is expected. 

 

 

Here's what I think is going on, and take the SP2-12000 for example:

 

Brian lists the voltage output at 126Vrms, which equates to almost exactly 2,000w at 8ohm (rated power spec for 8ohm).  I was able to squeeze 127Vrms at 40hz (5.5ohm)and 131Vrms out of that amp at 20hz (4.5ohm).  BOTH of those voltages exceed the spec sheet of that amp, which would also explain the distortion I saw in the wave form. 

 

With the XLS, and without any subs connected to the amp, it hits the brakes at ~100-102Vrms.  According to Crown's specs, it should be able to reach 111Vrms without any load connected to meet specs (rated at 1,550w at 8ohm).  However, my testing lasts for several seconds, so I'm not surprised the Crown falls a bit short (It was only $299 new!!).

 

I'm going to re-test with the Fluke 87V, but additionally, I'm going to list the maximum Vrms the amps will produce at 40/20/10/5hz without any subs connected first.  After that, then I'll test with a load connected. 

 

And yes, my impedance charts are done 1hz at a time with me imputing the readings into a spreadsheet...I should just buy DATS already!

[dgage]

And yes, my impedance charts are done 1hz at a time with me imputing the readings into a spreadsheet...I should just buy DATS already!

Let me get this straight, you just bought a Fluke 87V to go with your Fluke 115 and Fluke 117 but haven't ponied up for a DATS system yet? Really?

[Ricci]

Let me get this straight, you just bought a Fluke 87V to go with your Fluke 115 and Fluke 117 but haven't ponied up for a DATS system yet? Really?

 

Bwahaha! Exactly.

 

What the hell Luke? I think this might explain some things. Spend $150 and get a Smith & Larson WT2 if you want an easy button. I would pay the extra for this over the DATS or WT3. I've had all 3. Or save some money and wire up a voltage divider network like figure 2 in the attached pic and use REW or go ahead and get the ARTA suite for LIMP for next to nothing. Either way you shouldn't be manually plotting impedance 1Hz at a time in 2016.

 

[lukeamdman]

Let me get this straight, you just bought a Fluke 87V to go with your Fluke 115 and Fluke 117 but haven't ponied up for a DATS system yet? Really?

 

 

Bwahaha! Exactly.

 

What the hell Luke? I think this might explain some things. Spend $150 and get a Smith & Larson WT2 if you want an easy button. I would pay the extra for this over the DATS or WT3. I've had all 3. Or save some money and wire up a voltage divider network like figure 2 in the attached pic and use REW or go ahead and get the ARTA suite for LIMP for next to nothing. Either way you shouldn't be manually plotting impedance 1Hz at a time in 2016.

 

LIMP Voltage Divider.jpg

 

[lukeamdman]

Before I even start anymore testing I'm going to get my 87 calibrated.  I don't think it's off or anything, but it'd be nice to have the peace of mind that the numbers are accurate. 

 

Lucky for me I have a buddy who works at a machine shop, and he used to work for the calibration company that tests/calibrates all their tools!

[lilmike]

Bwahaha! Exactly.

 

What the hell Luke? I think this might explain some things. Spend $150 and get a Smith & Larson WT2 if you want an easy button. I would pay the extra for this over the DATS or WT3. I've had all 3. Or save some money and wire up a voltage divider network like figure 2 in the attached pic and use REW or go ahead and get the ARTA suite for LIMP for next to nothing. Either way you shouldn't be manually plotting impedance 1Hz at a time in 2016.

 

LIMP Voltage Divider.jpg

 

Um, what he said. Really.

 

Man, with the rest of the gear you have, skip the DATS, spend a little more and get something that works. 

 

http://www.woofertester.com/

 

$160. Plug and play, solid performance, and accurate. 

 

I have all kinds of ways to measure impedance. DATS, WT2, LIMP, SoundEasy, and REW, to start. The DATS collects dust, and pretty much has since I purchased it. I had the O.G. Peak Instrument Woofer Tester way back when and used that for YEARS. Only reason I got the WT2 was the desire to have something that worked in Windows. LIMP came along because ARTA does things the WT2 can't, and I needed to go there. SoundEasy? A friend strongly recommended it, I'm still learning what I can do with it. For checking cabinets though? The WT2 is the first thing that gets connected after my ohmmeter. 

 

If I didn't have the WT2, I'd use ARTA, SoundEasy, or REW with a simple homemade test jig. DATS just doesn't work for me anymore.

 

DATS sorta worked with W7, but it was squirrley. Now that I'm on W10, I can't get the DATS to calibrate properly or even measure the same driver (or resistor) repeatably. I'm not impressed with it at all. 

My settings are correct. I've tried updating to the latest software (for even more $$$), still no joy. At this point, I'm done wasting time and money on it. 

 

The DATS was an impulse buy from PE several Black Fridays ago, and I have pretty much regretted it since. Basically, mine is a useless piece of junk. 

[3ll3d00d]

DATS sorta worked with W7, but it was squirrley. Now that I'm on W10, I can't get the DATS to calibrate properly or even measure the same driver (or resistor) repeatably. I'm not impressed with it at all. 

My settings are correct. I've tried updating to the latest software (for even more $$$), still no joy. At this point, I'm done wasting time and money on it. 

 

FWIW (probably not much as you have multiple other ways to measure!) there was a thread on techtalk about that, various users had similar problems but settings were worked out that fixed the problems. 

[lilmike]

Oh, I tried them. I did my homework. I have exchanged emails with Mulcahy and the PE tech guys. I really wanted mine to work, because when it does, it is quick and easy. I do know a fair bit about setting up measurement gear and working with Windows audio. I've used lots of different gear. DATS doesn't measure up. 

 

There is a reason that TrueAudio/Dayton revised the DATS hardware as well as the software (V2 isn't the same as V1). They didn't do it for grins, there are real stability problems with V1. Mine has become a freaking random number generator. I talked to PE about it, all they want to do is sell me another DATS at MSRP. Thanks guys, I'll pass on that. I'd roll those dice at $40, not $100. I already bought the software upgrade...didn't help.

[SME]

I bought a WT2 to use for my builds.  It does have a pretty clunky 1990s interface that runs only on Windows, but it did basically work out of the box without much fiddling.  I found it very helpful to just hook it up and go when I needed to verify my recent speaker builds.  I would recommend it to him to at least get a good grasp of the parameters of his subs.  It would also be useful to observe any changes occur over the course of break-in.

[lukeamdman]

ok ok...I'll buy WT2 already!  Jeepers  

[lilmike]

ok ok...I'll buy WT2 already!  Jeepers  

 

That or just make a simple measurement jig and use REW.

 

There's no reason to measure impedance manually any more, there just isn't. Even if you do that sort of thing for fun.

[dgage]

Guess I'll need to order the WT2 to go with or replace DATS V2.  They really need to update the website as I looked at Woofer Tester but passed on it due to the old website and wondering if it was still supported.  Good to know it is good, wish I would have known 6 months ago.

[lilmike]

Guess I'll need to order the WT2 to go with or replace DATS V2.  They really need to update the website as I looked at Woofer Tester but passed on it due to the old website and wondering if it was still supported.  Good to know it is good, wish I would have known 6 months ago.

If your DATS works, by all means, use it.

 

Mine has become a random number generator, and nothing I've tried to date has fixed it.

[dgage]

If your DATS works, by all means, use it.

 

Mine has become a random number generator, and nothing I've tried to date has fixed it.

 

One thing I've learned from Mr. Ricci is to have redundancy in measurements, whether that be two mics, two SPL generators, or two meters.  So I'll use the DATS and Woofer Tester and hopefully they will correlate well.  If they correlate and measure similarly then I'm good and will have confidence in the measurements.  If they differ then I'll have to shoot one of them.  Hmmm, reminds me that I haven't been to the range in a while.

 

And you do know a random number generator isn't truly random right?  So if your DATS is truly random, then you might have something valuable on your hands. LOL!

[SpeakerPower]

Confucious say, "Man with two watches never know what time it is"

[dgage]

Confucious say, "Man with two watches never know what time it is"

 

LOL!  Good one!

[Stormwind13]

Dropped my peavey cs-4080hz off for testing. Will see what happens!

[andy497]

I think I might be having an issue with my scope...

 

Using a pair of multi-meters it all works fine.

 

Using the scope, a 10x probe on channel 1 to measure Vrms directly out of the amp also works fine.  However, as soon as I even connect probe #2, the readings for probe #1 are severely reduced. 

 

 

The two probes seem to interfere with each other, and somehow it's actually causing current to run through the probes and heat them up!?!?!

 

It's part of the design of the scope to have ground terminals of all channels wired together and to the main earth ground.  It would be unsafe not to.

 

In the above image, you can see that the black side of probe 2 is on the other side of the load from the black of probe 1, which is your problem.  There is huge potential between those two when they are on either side of the measured load.  As to why this happens, the resistance on a single probe from + to - (where the scope measures it) should be around 1 megaohm, so it can measure high voltages without things smoking, but the grounds between two probes on separate channels are probably < 1 ohm (you can continuity from the outside of two BNC barrels on the scope with your fluke to confirm).

 

In the image below, I reversed where you're clipping red and black on probe 1.  I think that should keep your scope happy with the grounds always at the same potential (but I would carefully test starting at low output; I am not an EE), and you'll be able to capture live real-time data the way the good lord intended.  You can also click Invert on any channel in the rigol to turn things back right-side up.

 

While you're doing that, it would be neat to see THD at various levels from the amp.  I know you can hit MATH -> FFT to view the full harmonics of a signal in db scale below the main graph, but I'm not sure if it spits out THD measurements.  Probably.  In any case, this setup would be worlds neater and easier than copying down fluke readings with pencil and paper.

[lukeamdman]

It's part of the design of the scope to have ground terminals of all channels wired together and to the main earth ground.  It would be unsafe not to.

 

In the above image, you can see that the black side of probe 2 is on the other side of the load from the black of probe 1, which is your problem.  There is huge potential between those two when they are on either side of the measured load.  As to why this happens, the resistance on a single probe from + to - (where the scope measures it) should be around 1 megaohm, so it can measure high voltages without things smoking, but the grounds between two probes on separate channels are probably < 1 ohm (you can continuity from the outside of two BNC barrels on the scope with your fluke to confirm).

 

In the image below, I reversed where you're clipping red and black on probe 1.  I think that should keep your scope happy with the grounds always at the same potential (but I would carefully test starting at low output; I am not an EE), and you'll be able to capture live real-time data the way the good lord intended.  You can also click Invert on any channel in the rigol to turn things back right-side up.

 

While you're doing that, it would be neat to see THD at various levels from the amp.  I know you can hit MATH -> FFT to view the full harmonics of a signal in db scale below the main graph, but I'm not sure if it spits out THD measurements.  Probably.  In any case, this setup would be worlds neater and easier than copying down fluke readings with pencil and paper.

 

I did successfully test the rig with the resistor on the the ground side, but I ran into the issue again with bridged amps since in that configuration there isn't a "ground level" connection...

 

Besides purchasing a very expensive low voltage differential probe to overcome that, I'll have to stick with the DMMs.  The 87V is getting calibrated this week.