Here is a zip file with both cases. I, really, would like to find what parameter/parameters simulate with so big difference and what model is closer to real transistors, Cordell's or ROHM's.
Damir
Damir
I thinked it is VAF param between QEMZ2 NPN & PNP (114.03 & 50.700). BC550C, BC560C are 162 and 160.
If I changed this value in rohmmodels.txt to make them equal -> THD decrease.
Thanks ilkafrv, it halved distortion, but must be more parameters, as with Cordell's models distortion was ten time lower. I curious how accurate are Cordell's models. It is not so important as distortion with ROHM models is very low too, more then 120 dB at 20 kHz..
Vaf parameter represents Forward early voltage, so with a higher value the distortion is lower due to less early voltage distortion.
Yes, I know that, but it is not whole story. I tried with onsemi models for bc846bpdw1t1g(NPN part VAF = 21.7, PNP part VAF = 105) changing VAF to 160 but distortion stayed the same. I am not sure how good are those company models, and I hope that Cordell's are closer to real thing. Onsemi does not have models for BC850C and BC860C nor for TO92 type(BC550/560) so I can't comparethose with the Cordell's. I hope that onsemi is using the same chip in both type TO92 and SMD.
The lower distortion is dependent on the frequency, early voltage wont affect frequencies above 1 to 3 kHz much, if at all.
for my personal experience Cordell models normally give less distortion than others models, and in some cases the early voltage is to high like in the mje15032, in the first picture is the curve trace of the onsemi model of the mje15032, and in second the cordell model. As one can see there are huge differences, and using the cordell model give much less distortion.
Vaf in cordell model is 2000+ and in onsemi model is only 31,5.
Vaf in cordell model is 2000+ and in onsemi model is only 31,5.
Attachments
Damir, the matching of the parameters from pnp and npn also play a part, if you look at the cordell models youll see the VAFs are near identical. I highly doubt this will be the case in actual part as is the case of the cob being the same.
A difference between the pnp and npn parts will bring about vas fighting.
Those rohm parts seem to be as good as the bc so in actual build the performance will be very much the same. The BCs might have a little lower noise.
A difference between the pnp and npn parts will bring about vas fighting.
Those rohm parts seem to be as good as the bc so in actual build the performance will be very much the same. The BCs might have a little lower noise.
Sergio, the mj15032 will have a very high VAF. Its rated at 250 v and 8 amps.
Someone at onsemi must be confused, VAF of 31.5. 🙁
In this case Cordell is correct.
Thank you all for your help.
I tried to increase BF for ROHM EMZ2 to match that of BC550C, but it did not influence distortion simulation.
I tried to increase BF for ROHM EMZ2 to match that of BC550C, but it did not influence distortion simulation.
If I change VAF of QEMZ2 to 162 / 160 like BC550C / BC560C then I got:
Total Harmonic Distortion: 0.000007%
So big and equal in VAF parameter has great influence to THD in this case.
Total Harmonic Distortion: 0.000007%
So big and equal in VAF parameter has great influence to THD in this case.
Is this THD level at 1 khz or 20 khz ??
Check the 2. Qemz2 has slightly lower cob if I remember correctly so may have advantage at 20 khz.
Thank you all to your help. I was one week out of internet reach, and I have to check that again.
Damier, do you still have some boards ??
I will soon have some time that I could use for a build. There is strong competition though, Ive been messing with Tpa6120 and AD815 type ICs for commercial design and they are extremely good, in fact Im very surprised at the performance these parts provide.
I like your design because of the No NFB though and Im very curious on performance comparison.
I will soon have some time that I could use for a build. There is strong competition though, Ive been messing with Tpa6120 and AD815 type ICs for commercial design and they are extremely good, in fact Im very surprised at the performance these parts provide.
I like your design because of the No NFB though and Im very curious on performance comparison.
Manso, I've sent you email, tell me if this is OK with you? I don't have any spare board any more.
Damir
Hi,
Finally I'm building the Gainwire mk2! 😉
I have a S2 Series Cliff jack socket (S2/BBB, 3P, 6 pins) and I would like to switch from loudspeakers to headphone listening (and the opposite).
How to do that ?
Finally I'm building the Gainwire mk2! 😉
I have a S2 Series Cliff jack socket (S2/BBB, 3P, 6 pins) and I would like to switch from loudspeakers to headphone listening (and the opposite).
How to do that ?
- Status
- Not open for further replies.