18 minutes ago, Pars said:

A1943/C5200

Same Here,also uses MJL21193/94 on other builds

Thanks guys. That is a good steer to transistors that I can buy! What about the driver transistors? I have a load of MJE340/350 that might fit the bill. Also very linear with flat beta vs Ic? Any thoughts?

OK - I've been through the whole thread trying to answer the question of why tantalum bead caps for the four 47u (C7, C9, C13, C15)  and the two 10u caps (C3, C16)? I failed in the quest.

So - what is the reason for using tantalum bead rather than aluminium electrolytic in those locations?

10 minutes ago, Craig Sawyers said:

OK - I've been through the whole thread trying to answer the question of why tantalum bead caps for the four 47u (C7, C9, C13, C15)  and the two 10u caps (C3, C16)? I failed in the quest.

So - what is the reason for using tantalum bead rather than aluminium electrolytic in those locations?

I think tants typically will have lower effective series resistance at high frequencies than electros, and in general tants esr decreases with increasing frequency unlike electrolytics. 

https://www.doeeet.com/content/eee-components/passives/why-low-esr-matters-in-capacitor-design/

Edited November 5, 20232 yr by jamesmking

Time to power up my HP 4275A LCR meter https://www.testequipmenthq.com/datasheets/Keysight-4275A-Datasheet.pdf and put this to the test 😉

3 hours ago, Craig Sawyers said:

Time to power up my HP 4275A LCR meter https://www.testequipmenthq.com/datasheets/Keysight-4275A-Datasheet.pdf and put this to the test 😉

I just have a Mastech MS5308 Lcr-Meter. 

kemet 35v 47uf tantalum esr

100hz 0.4ohm

120hz 0.4ohm

1khz 0.22ohm

10khz 0.19ohm

100khz 0.12ohm

 

panasonic 47uf 63V electrolytic esr 

100hz 1.1ohm

120hz 1.0ohm

1khz 0.39ohm

10khz 0.33ohm

100khz 0.30ohm

 

chinesium "suntan" brand electrolytic 47uf 105C 25V esr

100hz 4.8ohm

120hz 4.4ohm

1khz 1.75ohm

10khz 1.41ohm

100khz 1.26ohm

 

so for this tiny sample size (and not the same voltage rating) tants have lower esr throughout the range I can measure. so not enough data points, resolution, sample size or frequency range etc. to make any real conclusions but maybe enough to start a electro vs tant thread... 

 

 

 

 

Edited November 5, 20232 yr by jamesmking

On the basis of your measurements for PSU applications perhaps they have the edge. But for audio signal use they are absolutely lousy.

It is latish here in the UK so I'm not going to haul my HP gear out. But I'll try some stuff tomorrow.

QuantAsylum QA403 tests

Tested my GRLV noise level with the QA403 by connecting it directly with BNC test leads.
The maximum DC input of the QA403 is 40 VDC.
Measured the positive and negative voltages separately to see each noise level.
Used the left positive input BNC and put a terminator on the negative input BNC.
Installed software from https://github.com/QuantAsylum/QA40x/releases/download/1.201/setup_QA40x_1.201.exe
Start app and Select File-> Devices-> QA403
Select File-> New Settings, Set the FFT size to 128k, the full scale input to 0 dBV, and hit Run.
The first image is with the 36" BNC test leads alligator clips shorted together.
The second image is the QA403 connected to the GRLV with no load.
The third image is the GRLV with CFA2 as the load.
Have less than 8 uV RMS Volts on each output.

Was able to run RMAA on the QA403 by installing ASIO drivers.
https://github.com/dechamps/ASIO401
Install https://github.com/dechamps/ASIO401/releases/download/asio401-2.0/ASIO401-2.0.exe
Have to adjust the input/output scaling with the configuration file ASIO401.toml in
the Windows user profile folder; for example: C:\Users\Your Name\ASIO401.toml
Make sure the file doesn't have .txt on the end of it.
My settings for the RMAA loopback cable tests are:
fullScaleInputLevelDBV = +6.0
fullScaleOutputLevelDBV = -2.0
Adjust levels when testing an amplifier with gain.

The input values can be: 0.0, +6.0, +12.0, +18.0, +24.0, +30.0, +36.0 and +42.0
The output values can be: -12.0, -2.0, +8.0 and +18.0

Install RMAA https://audio.rightmark.org/downloads/rmaa6.exe
Right click and Run as administrator.
Start it and select ASIO401 for input and output device.
Select 24 bit and 48 kHz.
‌Install four coax bnc loopback cables from output to input on the QA403.
select Playback/Recording icon.
start test if signal levels are good, otherwise adjust levels in ASIO configuration file.
when test is done, save to location.
Generate a RMAA test report with the paperclip icon.

The test report has misspelled configuration signal as singal.
I used hexinator app to edit the rmaa6 executable C:\Program Files (x86)\Rightmark\rmaa6
Select Edit, Find, singal at 496C00
Change the hex code 6E 67 to 67 6E and save.

Here are the results of the RMAA balanced loopback tests in the last image.
A picture of the QA493, GRLV, CFA2 is the fourth picture.
The GRLV has the transformers in the rear box and the regulators in the front box.
Two steel panels separate them to lower noise.

 

Edited September 15, 20241 yr by dip16amp

might want to try again with a good quality mylar cap say 1uf in series with the power supply outout. you should be able to get better resolution.

8 hours ago, kevin gilmore said:

might want to try again with a good quality mylar cap say 1uf in series with the power supply outout. you should be able to get better resolution.

I put a solen mkp 1uf between the QA403 input and GRLV output and got more noise.

 

much better. max sensitivity of the qa403. 60 hz and multiples are very hard to remove, moving wires around in the box and things will change.  completely inaudible.   the synchronous rectifiers may take out even more noise. mumetal shielding around the transformer is also a good idea.

glad that more and more people are buying the qa403.  higher quality diy coming soon

 

Here is the A-Weighted measurement.

 

Here is the QA403 with BNC to XLR cables run on my balanced CFA2 amplifier with RMAA.

The ASIO401 driver is configured for -12.0 output to adjust for the 10 dB gain of the amplifier.

 

Here is the QA403 software for comparison to RMAA.

First image is noise.

Second image is THD & THD+N.

 

 

Edited September 19, 20241 yr by dip16amp