If you are testing with the same Ic and Vce the heat should be the same. Hfe 50 sounds ok but 150-220 is out of spec. Check your measurements.
This is how I did my measurements: for each device I adjusted voltage to 2V across the 1R22 collector resistor. After 15sec the devices were so hot that it was barely just possible to touch them by hand. Vbe was at 0.65V for all devices, HFE measurement was taken across the 100R base resistor and calculated with the formula hFE=(VRc/Rc)*(Rb/VRb). The calculatings for the Sescosems and Thomsons gave around HFE=50, for the IRFs and STC 150-220: the double die devices read around 3.2V across the base resistor, the single dies around 0.9 Volts.
Am I wrong?
I just rewired my test rig and this time took current readings directly from C and B of the device under test - same results...
I don't know how I could evaluate hFE in common-collector configuration: would you be so kind and point me to a source I can look into?
I again checked my calculations and measurements. I expected lower hFE with higher collector current, but it seems I get the opposite. Here my readings of the (single die) device with the weakest (a) and the strongest (b) hFE:
a) with Ic=0.4A I get hFE=80, at Ic=1.7A I get hFE=160
b) with Ic=0.4A I get hFE=140, at Ic=1.7A I get hFE=230
Is that normal?
a) with Ic=0.4A I get hFE=80, at Ic=1.7A I get hFE=160
b) with Ic=0.4A I get hFE=140, at Ic=1.7A I get hFE=230
Is that normal?
i am having a problem with a slow oscillation when i connect the output transistors to the amp boards. 2 times a second (2hz). it loads and discharges. i'm on dim bulb, and varian. the oscillation starts when the voltage goes over +-8V. When i measure the voltages, and resistances I'm not seeing anything unusual until it starts loading and dumping.
also i am not exactly certain how the ground pins are supposed to be routed. There are two on each board that don't share continuity.
it's possible the problem is regulation related, but the bias shoots up from 2mv to 50mv with each pulse, which makes me think drivers are not happy somehow.
If anyone who has experience with that symptom has advice i would appreciate it.
also i am not exactly certain how the ground pins are supposed to be routed. There are two on each board that don't share continuity.
it's possible the problem is regulation related, but the bias shoots up from 2mv to 50mv with each pulse, which makes me think drivers are not happy somehow.
If anyone who has experience with that symptom has advice i would appreciate it.
Yeah that's also my experience: with these boards and normal speaker cable it was impossible to get a clean output, but as soon as I tried the right cable it was fine. BTW Jeff, why did you change the zobel values...?
> BTW Jeff, why did you change the zobel values...?
I was trying to figure that out just the other day (while doing the Zobel for my JamJar amp). Sadly, I can't remember.
FWIW, I ended up going with 100n/10R for JamJar. But there doesn't seem to be much in the way of science around it and it appears that anything around those values works.
I was trying to figure that out just the other day (while doing the Zobel for my JamJar amp). Sadly, I can't remember.
FWIW, I ended up going with 100n/10R for JamJar. But there doesn't seem to be much in the way of science around it and it appears that anything around those values works.
well i'm trying to power up to set bias and measure offset, with no input or load on output. but you suggest running a speaker wire off the positive output? ok easy enough. i'll do that next.
can you explain where to route the ground from reg board, the pin between the regulated outputs. and where to route the "ground reference" that is next to the speaker out pin.
can you explain where to route the ground from reg board, the pin between the regulated outputs. and where to route the "ground reference" that is next to the speaker out pin.
I had exactly the same problem with NCC200 clone boards. Dim bulb tester was like a disco light. After a week or so of trying I narrowed the problem to wrong resister value. See attached picture and double check the resistors in this area, it may help you.
Best Wishes
Attachments
The cable needs to be the right sort, when I tried cardas speaker cable it didn't work. Try a minimum of 3.5m of Naim NACA4 or NACA5 or Linn K20 and connect both hot and ground to it and preferabely some test speakers!
Today I tried various output devices with one of Jeffs boards:
All except the MJ15003s and ST BUV20s had a bad clipping like distortion:
The ST BUV20s are probably fakes, see earlier in this thread: at that time they sounded much better than the MJ15003s, but today it was the opposite: I had 8R2 and 220n (Wima MKS) in the Zobel and can say that this sounds way more Naim like than the previous 20R / 100n I had in place. The MJ15003s sounded thin in the bass and sharp in the top end, now with 8R2 / 220n everything sounds as it should.
Makes me wonder if Naim "customizes" the sound signature of their product range according to the price tag using different Zobel values...?
Also I tried with and without the CCS mod from the HackerNAP (split the 22k near TR3 to have 10k + 12k and connect the junction via a 3u3 film to -ve, with the 12k resistor heading towards ground): it was a nice improvement indeed, I will keep the mod in place!
My conclusion: forget the vintage trannies, use MJ15003s, the zobel makes way more difference, and consider the CCS mod...
All except the MJ15003s and ST BUV20s had a bad clipping like distortion:
The ST BUV20s are probably fakes, see earlier in this thread: at that time they sounded much better than the MJ15003s, but today it was the opposite: I had 8R2 and 220n (Wima MKS) in the Zobel and can say that this sounds way more Naim like than the previous 20R / 100n I had in place. The MJ15003s sounded thin in the bass and sharp in the top end, now with 8R2 / 220n everything sounds as it should.
Makes me wonder if Naim "customizes" the sound signature of their product range according to the price tag using different Zobel values...?
Also I tried with and without the CCS mod from the HackerNAP (split the 22k near TR3 to have 10k + 12k and connect the junction via a 3u3 film to -ve, with the 12k resistor heading towards ground): it was a nice improvement indeed, I will keep the mod in place!
My conclusion: forget the vintage trannies, use MJ15003s, the zobel makes way more difference, and consider the CCS mod...
Attachments
