ETI 477 article

Some more measurements:

THD+N, 1KHz into 8 Ohms (80KHz bandwidth):

100mW: 0.0318%
200mW: 0.0214%
500mW: 0.0146%
1W: 0.0116%
2W: 0.0094%
5W: 0.0081%
10W: 0.0074%
20W: 0.0067%
50W: 0.0065%
100W: 0.0073%

I didn't go beyond 100W, as the heatsink was getting fairly hot, and I didn't want to cause any damage. I don't think the level of heatsinking it's got is really adequate for 100W, especially as I was only driving one channel.
 
ETI March '81, page 37 has a plot of distortion versus frequency. It claims the THD is a bit below 0.001%, at 1KHz, 10W.

My SMD version of the 6000 does 0.0013% THD, putting 100W into 8 Ohms at 1KHz (using the same generator and THD analyser), so it's about 10-15dB better, as shown in the following plot:

An externally hosted image should be here but it was not working when we last tested it.


I didn't have much time to measure it today. I'll repeat the exercise tomorrow at 10KHz, and do a frequency sweep at 100W as well.

I don't suppose anyone has an original 6000 that I can borrow to measure?
 
More numbers.

10KHz THD: (8 Ohm load, 80KHz bandwidth)

10mW: 0.0681%
20mW: 0.0613%
50mW: 0.0507%
100mW: 0.0382%
200mW: 0.0298%
500mW: 0.0265%
1W: 0.0278%
2W: 0.0294%
5W: 0.0303%
10W: 0.0305%
20W: 0.0306%
50W: 0.0300%
100W: 0.0530%

Frequency sweep: (50W, 8 Ohm load, 80KHz bandwidth)

100Hz: 0.0090%
200Hz: 0.0088%
500Hz: 0.0060%
1KHz: 0.0068%
2KHz: 0.0100%
5KHz: 0.0190%
10KHz: 0.289%
20KHz: 0.0205%
 
Alas none of our spectrum analysers go low enough in frequency.

fayef, I've had some luck diagnosing the problems with the right channel. It appears one of the output FETs has gone - a 2SJ49. If you look at the photo you posted up, you can see one of them has been running really hot, so has corroded more than the others.

It's also been oscillating, and has cooked the resistor in the HF shunt.
 
ETI477 article

Suzy
I noticed in the photo that the 4R7 had overheated.
BTW, with the AEM 6000, years ago, I noticed the occasional edginess on one channel only. Eventuually, I found that one pair of Mosfets was not current sharing very well.Just replacing the one that wasn't drawing nearly as much current completely cured the problem.It wasn't faulty.

SandyK
 
feyef
It will need more than the 2SJ49 to fix. All the electros except the PSU electros will need replacing. The 4R7 that is discoloured in the photo will also need replacing. May also need to address the cause of the reason for the instability that burnt the 4R7.
However, I am certain that there was no HF oscillation problem originally, as it always paid to check that those resistors only get warm when HF sine wave testing,and NOT during normal operation.
Perhaps it may be prudent to add a small amount of G to S capacitance to the 2SJ49s ?
Let's see what Suzy recommends.
SandyK
 
Suzy,

My old 6000 is definitely available if you want to borrow it for testing as a control. I fired it up recently and it seems to work quite OK except that there's a slight crackle that comes and goes in one channel. I suspect one or more electrolytic caps need replacing, but this should not impact too much on test results.

Let me know if you want me to deliver it with my company awd gas guzzler. I live at an inconvenient arm of the Upper North Shore which is probably a few potato scallops over the usual fish and chip distance from where you live.

Sandy K, (and others)

While I think of it, I found a sealed package of Intersil dual jfets
in one of my old component boxes. These are the 20 yo devices that were originally specified by DT for the 6000 / 6005 modules. There are 6 of them so if anyone desperately needs a pair I might be able to part with 2x.

Cheers,
vm
 
VHF man said:
Suzy,

My old 6000 is definitely available if you want to borrow it for testing as a control. I fired it up recently and it seems to work quite OK except that there's a slight crackle that comes and goes in one channel. I suspect one or more electrolytic caps need replacing, but this should not impact too much on test results.

Let me know if you want me to deliver it with my company awd gas guzzler. I live at an inconvenient arm of the Upper North Shore which is probably a few potato scallops over the usual fish and chip distance from where you live.


It's in the name of science, so I'm sure the environment won't mind.

I'll PM you my address - you can have a listen to my SMD version too :)

Getting back to the 5000, I had a bit of a look at it today. It's being a very naughty amp - I removed the Zobel resistor (which had completely fried, leaving a burnt patch on the board beneath), powered it up with my dummy load connected and got a really beautiful oscillation out of it on the positive half cycle, of maybe 1V, superimposed on the 1KHz test tone I was putting through it.

But by the time I got the cro set up to take a happy snap, it'd gone :( Poking and prodding completely failed to bring it back. Anyway, I've attached a pic showing the output distortion, which is pretty impressive.

I didn't make terribly much headway with it at lunch today - as I spent most of the time checking out the mods to Tilbrook's original design. The zeners to balance the dissipation in the two sides of the VAS are a very good idea.

Anyway, I've ruled out blown output transistors - they're all sharing current happily. I suspect things will look a little better once the electrolytic cap in the feedback path is replaced - it's bypassed with a poly cap, and I suspect the poly cap is providing all the capacitance.
 

Attachments

  • lecroy.png
    lecroy.png
    34.8 KB · Views: 776
ETI 477

Suzy
David Tilbrook agreed with my mods when he met a few enthusiasts at Jaycar's Gore Hill store one Saturday morning.
Im glad to hear that the O/P devices are current sharing. I will be very interested to hear your views on how it sounds when you replace the electros.Hopefully it should sound a little less "interesting" next time.
Regards
SandyK
 
Modelling

Hi guys,

Just for fun, I've knocked together a simulation of the ETI477 in LTspice. I was able to find models for the BC550, BF469, and BF470 (included in the .zip). I substituted the 2SK1058 and 2SJ162 for the 2SK134 and 2SJ49, as I understand these are the same die in different cases.

The simulation agrees with the measured results reasonably well. 100W, 8 Ohm THD is 0.0042% at 1KHz, and 0.041% at 10KHz.

Interestingly, substituting KSA1391 and KSA3503 VAS transistors reduces THD to 0.00038% at 1KHz and 0.023% at 10KHz.
 

Attachments

  • eti477.zip
    13.1 KB · Views: 258
Oops. Sorry about that. Yes, Jaycee is right with the part numbers. I'm rather taken with these transistors for VAS use. They're lovely and fast, and readily available to boot (from Digikey).

It doesn't behave very well as all when clipping. Here's the result of a 4V 10KHz sinewave input.
 

Attachments

  • eti477_clipping.gif
    eti477_clipping.gif
    7 KB · Views: 711
Yes.. I have tried simulating more conventional amplifier designs (single transistor VAS) and they (or the models at least) seem to have a terrible time recovering from clipping. In a single transistor VAS scenario, they have a tendency to stick to the rail for a while.

This could be a simulator thing though. I'm using LTSPICE as well. I guess the only way to know for sure is to "suck it and see"!
 
ETI5000 Mods

I have seen the 2X 9V zener mods on the track picture but I can't quite see how the tracks have been cut are exactly where the zeners have been inserted in order to modify my old original ETI5000 for improvement since the picture resolution is too low.

In addition there also appears to be a cap connecting the source's of both 2SK134s in the mod. I would appreciate info on exactly how the mod is done.

Thanks !