Yipee !
Finally, i got some reasonable match for my drivers mje15030/1...
The result ? ~Factor 8 reduced thd...
Attached shot is from spectralab, same setup, still jfet version. (bjt is at least 2 times lower thd)
And: board layout not that bad...
This proves that the distortions i had in my measurings were caused by mismatch of these drivers.
That also means that close match of these drivers is mandatory for symasym...
BTW, i had checked minimum bias using MJL0281/0302... ~60ma (13mv across single RE) Bias below that showed higher order harmonics.
Tomorrow i have to find match for other channel and listen.
Mike
Pavel: Still "horrible" ?
Finally, i got some reasonable match for my drivers mje15030/1...
The result ? ~Factor 8 reduced thd...
Attached shot is from spectralab, same setup, still jfet version. (bjt is at least 2 times lower thd)
And: board layout not that bad...
This proves that the distortions i had in my measurings were caused by mismatch of these drivers.
That also means that close match of these drivers is mandatory for symasym...
BTW, i had checked minimum bias using MJL0281/0302... ~60ma (13mv across single RE) Bias below that showed higher order harmonics.
Tomorrow i have to find match for other channel and listen.
Mike
Pavel: Still "horrible" ?
Attachments
Hi Mike,
-Chris
Yup.
I've been saying that for years about most amplifiers. Every complimentary part should be matched, including current mirrors and diff pairs. It does make a difference. It's nice you were able to prove this. Something I knew but had little prove of except in gross cases.
Thank you. My setup is inadequate for high performance amplifiers. I really have to get my bench sorted out. I'll turn my bias up and have a more careful listen. Mind you, with my wiring spread all over, other problems may swamp the bias issues out.
-Chris
Chris, never mind, my wiring is horrible... I will get better values replacing that wire nest.
Gladly, symasym does not need complementary matches except the outputstage. The drivers matching is here most important as these are connected directly to the vas having quite high outputimpedance. Any asymetric current suckout from the drivers will heavily increase openloop distortion.
Mike
Gladly, symasym does not need complementary matches except the outputstage. The drivers matching is here most important as these are connected directly to the vas having quite high outputimpedance. Any asymetric current suckout from the drivers will heavily increase openloop distortion.
Mike
Hi Al, the measuring was all the time ~1W into 8ohms. Need to get some nice power resistors on a heatsink for extended measurings.
I was hunting the strange too high harmonics / THD. I always thought i get half waves from the power supply into the signal, but i was way wrong !
Anyway, this powerlevel is ~typical listening level, the amp got even hotter during measuring than at listening into 4ohms.
And, the measuring is definitely in ClassAB condition.
Mike
I was hunting the strange too high harmonics / THD. I always thought i get half waves from the power supply into the signal, but i was way wrong !
Anyway, this powerlevel is ~typical listening level, the amp got even hotter during measuring than at listening into 4ohms.
And, the measuring is definitely in ClassAB condition.
Mike
For the output devices the hfe value is the most important. Vbe is practically unimportant.
Unmatched bjts in the frontend will mainly cause unbalance/dc-offset, mismatch of the drivers mainly higher THD. You can easily check match of drivers by measuring the voltage drop across the basestoppers to the drivers.
Mike
Unmatched bjts in the frontend will mainly cause unbalance/dc-offset, mismatch of the drivers mainly higher THD. You can easily check match of drivers by measuring the voltage drop across the basestoppers to the drivers.
Mike
Hi Mike,
If you can get a hold of the Marantz 500 service manual you may find it illuminating. Keep in mind this amp was probably designed in 1968 for it's 1969 release. I think you will come away with a lot of respect for the design team. One important point lost an many is the input buffer. Newer Nakamichi receivers called it HTA. It presented the BJT diff pair with a low impedance source. (light should be coming on)
-Chris
I agree that's one very important thing. The reflected load should look relatively constant to the Vas. Of course this is not true in real life.
If you can get a hold of the Marantz 500 service manual you may find it illuminating. Keep in mind this amp was probably designed in 1968 for it's 1969 release. I think you will come away with a lot of respect for the design team. One important point lost an many is the input buffer. Newer Nakamichi receivers called it HTA. It presented the BJT diff pair with a low impedance source. (light should be coming on)
Crossover distortion is more easily seen at lowish powers, low impedance and higher frequencies. At higher power levels the effects of the Vas and other things become more visible.
Actually, I just go on hFE. Vbe differences will be small in comparison with other circuit imbalances. In the same batch of transistors the Vbe differences will be small anyway.
-Chris
Mike,
I have a huge problem...
ANY signal source I use in my sim (LTSpice) generates odd harmonics!
Does leaving the timestep blank make spice do an auto timestep? My computer isn't the fastest computer (Pentium 3 550Mhz, Win2k, 256Mb RAM), so I'm wondering if this has something to do with it. I have a screenshot:
I have a huge problem...
ANY signal source I use in my sim (LTSpice) generates odd harmonics!
Does leaving the timestep blank make spice do an auto timestep? My computer isn't the fastest computer (Pentium 3 550Mhz, Win2k, 256Mb RAM), so I'm wondering if this has something to do with it. I have a screenshot:
Attachments
Keantoken, i can't help you with that as i am not familiar with LT-Spice.
But for sure having an older computer does not generate different data here.
The timestep might need to be fixed (try ~200ns for example).
Just play around with the settings until you get a clean fft with just the signal source.
Mike
But for sure having an older computer does not generate different data here.
The timestep might need to be fixed (try ~200ns for example).
Just play around with the settings until you get a clean fft with just the signal source.
Mike
My damn cheap dmm can only measure millivolts, so its likely to be a much worse match than yours. I read 4mv on r8/9.
Having 60ma bias and 1.2ohm basestoppers and the MJLs hfe 100, i measure... zero... my damn dmm... (should be 0.7mv)
Hi Mike,
If you are going to buy a hand held meter, Fluke or nothing. A used Fluke is better than all others new. Trust me on that one. In order, 189, 87V, 179. They can be seen here.
I hope I got the correct spot, I don't speak German.
For a bench meter, Agilent. Period, end 'o story. Failing that, a Fluke bench meter. I bought an HP 34401A when they first came out. I've never regretted that purchase. My priorities may be different than your's though. The new ones are even better. Check out the AC frequency response on the Agilent. A used 34401A, plus a calibration will not steer you wrong.
-Chris
If you are going to buy a hand held meter, Fluke or nothing. A used Fluke is better than all others new. Trust me on that one. In order, 189, 87V, 179. They can be seen here.
I hope I got the correct spot, I don't speak German.
For a bench meter, Agilent. Period, end 'o story.
Failing that, a Fluke bench meter. I bought an HP 34401A when they first came out. I've never regretted that purchase. My priorities may be different than your's though. The new ones are even better. Check out the AC frequency response on the Agilent. A used 34401A, plus a calibration will not steer you wrong.-Chris
I have a Fluke DMM that measures capacitance (not as low as pF, though), resistance, diodes, frequency, voltage, and amperage. It's slightly old, but it works. I don't remember the model number because I would have to dig it up from the rubble that others call my room. I have way too much stuff... It's gonna all fall on me one of these days. 
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.