Yes, agreed. I actually am derating for 70+ C , no A/C ...hot 42C summer days , amps in clubs ,at parties. That is why I chose the 2sa1294/c3263 instead of the MJE's. Quite the overkill (below) , at 70C+ I still get 80W Pd. I sure in the auto industry , they did their homework - derating wise (much less current to negate the high temp) . With normal use , I want to feel almost NOTHING (finger test) , I know this will equate to "warm" out in the hot tennessee sun or in a closet at the club. 
OS
OS
Attachments
I got my amp repaired today, so its up and running again. I made some changes to the compensation and a few other areas which I will post up in the next few days.
By the way OS, I like your idea of using a hefty driver device rather than the 15032/33 approach. I will look into it in the next few days.
By the way OS, I like your idea of using a hefty driver device rather than the 15032/33 approach. I will look into it in the next few days.
I tried using the Solid State Relay (SSR), still with irf540, for an AC switch. I first tried without the two extra diodes and it allowed current to pass, but it did not cut off fully.
I modified the SSR by adding the two extra diodes. Again under test it allows current to pass, but does not cut off.
It appears the internal diodes in the irf540 and the external diodes are passing the current during the "cut-off" phase.
What am I doing wrong?
I modified the SSR by adding the two extra diodes. Again under test it allows current to pass, but does not cut off.
It appears the internal diodes in the irf540 and the external diodes are passing the current during the "cut-off" phase.
What am I doing wrong?
These are indeed hefty drivers. The 34/35 have an even higher SOA.
Only for very high output powers are higher SOA devices needed for drivers.
How are you driving the gates ? All the circuits I have seen for this use an isolated drive method... which Michael Bean mentions somewhere and his preferred method is photovoltaic couplers.
I have seen at least one design using a floating transformer to provide the bias... messy I thought.
Besides the slightly larger footprint , to-3p's are sometimes cheaper than to-220's , easier to mount , better thermal dissipation , SOA. T0-220's are a PIA with the insulated grommet - small size. TO-3p's , if used on a separate driver sink , will run much cooler .. as well.
Andrew is right , that much "grunt" is only needed for low Hfe outputs or many pairs , when I come back with my big stuff - a 5-6 BJT pair OP , the R/D will already be done.
PS - I pay $2.20usd for MJE15032 , $1.80 for NJW0281 (to-3p) , $2.02 for the TO-3p sankens.
OS
These devices have 250pF Cob , while drivers like 2SC4793 has about 20pF,
so it s not adequate to be driven by the BF469/470.
It would be more logical , in a triple EF scheme , to add darlingtons,
wich have internal low Cob driver, as final stages after the pre-drivers..
True that high quality darlingtons are pricey , but they exists...
Edit : One thing more...
High power devices such as the ones recomended by OS , have Ft
wich is optimum only at currents above 1A , while usual drivers will
exhibit optimum at currents of about 0.3/0.7A.
Last edited:
Andrew T,
I should have added, yes, they will conduct as any diode will, that's why they're connected with polarities opposed (they are in parallel with the internal substrate diodes that all MOSFETs have). The diodes are only there to protect the MOSFETs fron inductive kick-back, they aren't supposed to do anything unless there is a problem.
Mike
I should have added, yes, they will conduct as any diode will, that's why they're connected with polarities opposed (they are in parallel with the internal substrate diodes that all MOSFETs have). The diodes are only there to protect the MOSFETs fron inductive kick-back, they aren't supposed to do anything unless there is a problem.
Mike
If they are in inverse parallel it will always conduct... I'm trying to visualise what you are describing as your set up...
Have you tried like this,
http://www.diyaudio.com/forums/solid-state/155939-mosfet-relais.html#post2003541
http://www.diyaudio.com/forums/solid-state/155939-mosfet-relais.html#post2003541
FETs Source to Source gives success.
Combined Vdrop, of both FETs in series, changes from 0.4mVdc @ 10mAdc to 213mVdc @ 3.34Adc. ~40milliohms to ~64milliohms, seems a good result for a pair of <77milliohm FETs. At the higher currents I can clearly see the rising impedance with rising Tj.
The nice bit is that the effective impedance of the conducting FETs is lower at low current, i.e. during quiet music they have least effect.
Combined Vdrop, of both FETs in series, changes from 0.4mVdc @ 10mAdc to 213mVdc @ 3.34Adc. ~40milliohms to ~64milliohms, seems a good result for a pair of <77milliohm FETs. At the higher currents I can clearly see the rising impedance with rising Tj.
The nice bit is that the effective impedance of the conducting FETs is lower at low current, i.e. during quiet music they have least effect.
Last edited:
Andrew T,
I assume you're using the IRF540 that you mentioned before. Those devices are not the best choice for this application, Rds is too high, and you will see greater increase of Rds with increased current through them. There are many newer devices available with significantly lower Rds, they will work much better.
Mike
I assume you're using the IRF540 that you mentioned before. Those devices are not the best choice for this application, Rds is too high, and you will see greater increase of Rds with increased current through them. There are many newer devices available with significantly lower Rds, they will work much better.
Mike
Close to 500w into 4 ohms Andrew. That's a lot of power.
But the reason for using heftier drivers is that the driver stage can be run heavily into class A, which improves linearity. The open loop distortion of the driver stage when run at 100ma is about 40ppm in simulation driving a 15mA load (ie about 5k) at 60v peak (120 pk to pk). The 15032/33 are small and run close to their second breakdown , and especially so when the rIls can be as high as 77V which is what I measured last night.
Last edited:
500W into 4r0 is equivalent to 63.2Vpk and 15.8Apk into 4r0.
The loaded supply rails would likely be ~+-67Vdc.
The unloaded supply rails would likely be ~+-70Vdc.
Now looking at 4r0 resistive loading. A 5pair output stage.
15.8Apk divided between 5 devices gives an hFE ~60 @ 3.2Apk.
The base drive requirement would be ~ 15.8/60 ~ 260mA. For a driver stage biased to 100mA that is beyond ClassA for the drivers.
The drivers biased to 150mA on+-70Vdc supply rails will dissipate ~10W.
An MJE15034/5 can easily cope with that into 60degree phase angle loading for 500W into 4ohm speakers. Just fit decent sinks to the drivers.
100mA of Ib is nothing to these 50W devices.