Jump to content

Rob's Amp tests


peniku8

Recommended Posts

I investigated dielectric absorption quite a long time ago when I was 17 so my memory is a bit fuzzy.  I built a precision high dynamic range current mirror and monitored the charge/leakage current over the period of hours when the capacitor was connected to a voltage source with some series resistance.  I also had access to a volt meter with exceptionally high input impedance for monitoring the capacitor voltage.  I found that my capacitors fitted a model of a capacitor in parallel with another capacitor with a very high impedance in series with it that represented the long term leakage trend and the dielectric absorption. 

Static electricity can kill you if you use it to charge a large capacitor ;)

Link to comment
Share on other sites

  • 3 weeks later...

I somehow feel like this is bad design or something isn't working as it should. The 4 transistors I replaced in the PSU, which all broke after about 2 minutes of 4KW output generate a humongous amount of heat even in idle.

The heatsink seen below reaches 80°C (176°F) in idle and SOMEHOW I measure over 1KV from heatsink to ground and when I touch it with the thermo sensor I see small sparks. What is going on here??

fWzrDr9.jpg

 

uNp7W4e.jpg

 

Link to comment
Share on other sites

What confused me even more was the seemingly random choice of output transistors, as two different models were actually used in the amp.

One output channel has an even distribution of both types, while the other one has 5 of one type and 9 of the other. Both are the NPN type.

https://www.onsemi.com/pub/Collateral/NJW0281-D.PDF

https://www.onsemi.com/pub/Collateral/NJW21193-D.PDF

Looks like the 21193 is a bit better. I'll solder two from the broken pcb to the working pcb to "upgrade" the FP-13000 to Fp-14000 on the output stage. It might make a slight difference into lower impedances. 14 vs 16 transistors is still a decent upgrade.

Link to comment
Share on other sites

Seems like this amp is haunted by Hoffmann's Law. I'm not sure why but the modifications I made (wich was adding the two missing resistors and transistors) broke the amp stage. The fuses melted so I removed the components again but it's still inoperational. Welp, now the amp doesn't work at all anymore, since both output stages are broken. I'm guessing that it's a relatively minor thing, but I'm not the one to track down a broken component and I'm fairly certain that getting it fixed will be more expensive than the 150$ for the replacement board.

Link to comment
Share on other sites

This is so annoying. I was just way too curious to see if I could improve the amp's performance by adding those components and now I caused 150$ damage again.

Well, heres more random info on the PSU: when both amp stages are disconnected, the PSU hold over 200V for an hour. I used a light bulb to quickly drain it. When it's down to 0.1V, it takes about 10 seconds to climb back to 2V. I immediately connected the 230V light bulb after I powered the amp off, when the PSU was still at around 350V from positive to negative. It was a 30W light bulb or smth, it lit up brightly for a second and the light faded completely after some 20 seconds. Very convenient way when you're working on this kind of stuff.

  • Like 1
Link to comment
Share on other sites

The 1kv to ground sounds very weird as even if the heatsink was live on purpose you wouldn't expect 1kV to be generated anywhere in the amp.  Mix of output transistors is a very bad idea as they won't share current effectively.  On the plus side this is making me very certain I don't want one of these amps in my life :)

  • Like 2
Link to comment
Share on other sites

21 minutes ago, kipman725 said:

The 1kv to ground sounds very weird as even if the heatsink was live on purpose you wouldn't expect 1kV to be generated anywhere in the amp.  Mix of output transistors is a very bad idea as they won't share current effectively.  On the plus side this is making me very certain I don't want one of these amps in my life :)

I ordered two replacement modules, will see if the current ones are the same. This amp is 7 years old after all.

Link to comment
Share on other sites

Quick update on the Sanway: I've recieved two replacement amp boards they sent me for 300$ including shipping. Amp is operational again.
The amp boards which I had were version 2, the new ones are version 10. They now also use the same type of transistors throughout.

Link to comment
Share on other sites

This has all been quite interesting to follow.  I'm going to take a wild guess that the mixture of transistor types on the old board was "on purpose", albeit maybe as a last-minute hack or tweak for some other problem.  Hopefully the new revision is a lot more robust.

Link to comment
Share on other sites

15 minutes ago, SME said:

This has all been quite interesting to follow.  I'm going to take a wild guess that the mixture of transistor types on the old board was "on purpose", albeit maybe as a last-minute hack or tweak for some other problem.  Hopefully the new revision is a lot more robust.

Well the PSU blew up during testing. I only had to replace the amp boards because I damaged them by accident while trying to repair the PSU.

If the mix was on purpose, why use different amounts on both channels?

I also have the feeling that the amp gets hotter in idle now. So either the heat transfer from the transistors to the heat sink is better now or something else is the problem. I also reversed the fan mod I did and let the fans run full tilt now. I‘ll have a look at the idle power draw and see if it‘s higher than the ~150w it was before.

Either way, the bad gut feeling will now never subside when I bring that amp to a gig. I don‘t have any use for it at the moment anyways, so I might just sell it and call it a day.

Link to comment
Share on other sites

7 minutes ago, peniku8 said:

If the mix was on purpose, why use different amounts on both channels?

Re-work in QC?  I imagine that whatever is going on here, it's probably not good.  ;)

14 minutes ago, peniku8 said:

Either way, the bad gut feeling will now never subside when I bring that amp to a gig. I don‘t have any use for it at the moment anyways, so I might just sell it and call it a day.

I would agree unless you think you can develop your general knowledge/skills further by finding new ways to break it.  :)

Link to comment
Share on other sites

I measured the idle power draw and it looks like it went up to 160W (from 115). I can only measure in 0.1A steps, which is 23W, so the possible difference ranges from 23W to 69W. Not really a good sign, but it's not a huge amount either, so I'm not necessarily concerned about it. Their 4 channel amp draws almost 200W in idle. One reason for me to not switch the TSA1400 in the theater back out for the FP13000.

Link to comment
Share on other sites

  • 2 weeks later...
On 6/15/2020 at 7:29 AM, kipman725 said:

https://en.wikipedia.org/wiki/Dielectric_absorption

From my tests the absorption of electrolytics can approach 50% (much more than Wikipedia claims).  Solved via bleed resistors.  I had a college who liked to leave 5kV film capacitors the size of a thermos flask without bleed resistors attached on his desk and around the lab when I worked at the University.

On the subject of amps I see that people who have experience with the Lab clones are reporting better reliability with CVR amplifiers:

https://www.cvr-audio.com/product/101.html

 

 

 

 

I've decided to order one of the cvr d-3002's to see whats up with them. I don't have the equipment to bench test but I can do some a/b listening tests vs my K10. Ill report back

 

  • Like 2
Link to comment
Share on other sites

  • 5 months later...

I'll be posting some measurements of the Hypex FA123 soon. I'll also update my tests to streamline the process a bit.

What frequencies would ppl be most interested in? I thought of doing 1k, 60Hz for bass performance and 20Hz for sub-bass performance. I don't wanna take too many measurements, since everything multiplies so a 4 channel amp into 3 different loads with 3 different frequencies already makes 3^3=27 tests (1ch 2ch 4ch; 1khz 60hz 20hz; 8R 4R 2R).
I might as well drop the 1ch measurement for 4 channel amps as an edge case like that doesn't make much sense.

Link to comment
Share on other sites

The good thing about testing low power amps is that I don't have to fill the loadbank up with water.

Random facts:

  • It would take about 10 minutes to heat up the water in the tank to unsafe levels (beyond 60°C) at 3650W, which is the maximum power I can draw from the 230V 16A wall socket.
  • I'd need 500KW of power input to boil the entire body of water within a single 10 second test.
  • A full 10 second test at 5KW would heat up the water from 23°C (room temperature) to about 24°C.
  • Heating the water to 'lukewarm' would take about 5 minutes at full available 3,6KW power intput.
  • Running all heating elements at spec (230V or 3KW each; 48KW total), the water temperature would reach hot tub levels within half a minute.
Link to comment
Share on other sites

On 1/4/2021 at 5:26 PM, peniku8 said:

I'll be posting some measurements of the Hypex FA123 soon. I'll also update my tests to streamline the process a bit.

What frequencies would ppl be most interested in? I thought of doing 1k, 60Hz for bass performance and 20Hz for sub-bass performance. I don't wanna take too many measurements, since everything multiplies so a 4 channel amp into 3 different loads with 3 different frequencies already makes 3^3=27 tests (1ch 2ch 4ch; 1khz 60hz 20hz; 8R 4R 2R).
I might as well drop the 1ch measurement for 4 channel amps as an edge case like that doesn't make much sense.

Sounds like a good plan to me. I know this stuff isn't quick or easy so whatever makes sense to you...

Thanks for putting in the effort and sharing the results. 

Link to comment
Share on other sites

On 1/4/2021 at 5:44 PM, peniku8 said:

The good thing about testing low power amps is that I don't have to fill the loadbank up with water.

Random facts:

  • It would take about 10 minutes to heat up the water in the tank to unsafe levels (beyond 60°C) at 3650W, which is the maximum power I can draw from the 230V 16A wall socket.
  • I'd need 500KW of power input to boil the entire body of water within a single 10 second test.
  • A full 10 second test at 5KW would heat up the water from 23°C (room temperature) to about 24°C.
  • Heating the water to 'lukewarm' would take about 5 minutes at full available 3,6KW power intput.
  • Running all heating elements at spec (230V or 3KW each; 48KW total), the water temperature would reach hot tub levels within half a minute.

The resistance of the heating elements will change from the beginning to the test to the end of the test if you do not immerse them in water. This will change the results of the calculation of power into the load slightly.

Link to comment
Share on other sites

27 minutes ago, Ukko Kari said:

The resistance of the heating elements will change from the beginning to the test to the end of the test if you do not immerse them in water. This will change the results of the calculation of power into the load slightly.

That's correct, but they don't actually change by much, especially at only 50W per 3KW element (which is 100W into 8R=two elements in parallel). For 4R tests that's 4 elements in parallel already, so it's only about 30W per element on the FA123.

Link to comment
Share on other sites

  • 9 months later...

Been doing some tests with my Hypex amps and just out of curiosity I ran them without water in the tank, because I wanted to see how warm the elements got when testing lower power levels.

I put the bank on 8R (two elements in parallel) and gave it 50W for a minute. I could still touch the elements after the test was over. They were probably around 50°C (110°F?) at that point. Good news for me, as that means I can test everything below a few hundred watts without having to fill up the tank. I will give it a bit more power tomorrow to see how much the resistance changes, but it shouldn't be much. It's 8.2-8.3R cold and 8R under full load (3KW per element, which would mean an amp with 6KW into 8R would be required to saturate one channel).

Link to comment
Share on other sites

  • 2 weeks later...

I took some new distortion measurements of the interface and the Pmillet Sound Card Interface I'm using.
I decided to use REW's RTA to take these, as it gave it more flexibility with window sizes and averaging etc. 0dbV is not correct, but I had to 'calibrate' it to zero, since I couldn't find a way to take the fundamental as reference.
Thoughts?

wACkyST.jpg

 

JKC7RQK.jpg

Link to comment
Share on other sites

I did some loopback tests of a presonus firebox, they are ~7dB worse than your results for noise and distortion.  However your distortion is much lower.

https://www.audiosciencereview.com/forum/index.php?threads/presonus-firebox-24bit-96khz-loopback-measurments.24565/#post-828710

should be more than adequate for PA amplifier characterization.

Link to comment
Share on other sites

9 hours ago, kipman725 said:

I did some loopback tests of a presonus firebox, they are ~7dB worse than your results for noise and distortion.  However your distortion is much lower.

https://www.audiosciencereview.com/forum/index.php?threads/presonus-firebox-24bit-96khz-loopback-measurments.24565/#post-828710

should be more than adequate for PA amplifier characterization.

The Fireface had been RMA'd once and ever since it seems to reset the connection to the PC occasionally when a ceiling light (LED or Tungsten) is switched on or off, which is very annoying. I suppose there might be something wrong still, which might also cause the mains hum (50hz here in Germany) plus its 3rd harmonic, which is the reason the THD+N gets tanked. I think I should contact RME customer support to confirm whether these results are expected or not. I ran this through my patch bay in the outboard rack, which meant 50ft of cable, but it was balanced so it should not matter much?

Link to comment
Share on other sites

I figured out the mains hum, which was a rack power strip mounted above my patchbay in my outboard rack.
The following measurement was taken with the power strip switched off:

HEFPGle.jpg

THD looks good, but THD+N still pretty bad, I'm not sure why that is, as I can't really see anything that dirty in the rta there...

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...