Indeed, Q72 should be mounted on heatsink for thermal compensation; to use the HEC I would leave a simple resistor between the emiters of Q34,Q35 and implement the Zobel for each output device as indicated in Mr. Cordell's paper (close to the gate of each MOSFET).
I would experiment first without the mosfets mounted, to get ~2,5-4V between the gates needed to open these Toshiba devices.
Saying that, I am interested to see your results 🙂
I would experiment first without the mosfets mounted, to get ~2,5-4V between the gates needed to open these Toshiba devices.
Saying that, I am interested to see your results 🙂
Q72 & Q71 both mounted on heatsink, has to be the main heatsink or any heatsink (decent size) will do the job (in case if I remove the VBE).
Based on your advise take away the added VBE and put back the resisters how it was.
Zobel for output (pairs) or device that confusing to me, never saw something like that.
Can you make a sketch please so I can see what do you mean........
Thank you
Based on your advise take away the added VBE and put back the resisters how it was.
Zobel for output (pairs) or device that confusing to me, never saw something like that.
Can you make a sketch please so I can see what do you mean........
Thank you
It has to be the main heatsink, so they are monitoring the output devices temperature.
You can see the paper of Bob Cordell on his site:http://www.cordellaudio.com/papers/MOSFET_Power_Amp.pdf
The Zobels I refer to are depicted in figure 12, groups C8-R52 and C9-R53. For each MOSFET used, I would use such a Zobel group.
Cheers!
You can see the paper of Bob Cordell on his site:http://www.cordellaudio.com/papers/MOSFET_Power_Amp.pdf
The Zobels I refer to are depicted in figure 12, groups C8-R52 and C9-R53. For each MOSFET used, I would use such a Zobel group.
Cheers!
Thank you very much!
Now I see in real life how to apply that type of Zobel you refer or advise.
But do I have to go there
These circuit was built by Cordell himself with Hitachi mosfets. I assume you know about that.
The difference I want to use these Toshiba 2SK1529 & 2SJ200
The reason for changing the mosfet I have those at least 8 years and I never find a good circuit where I can use them.
OK I go back to the original circuit (remove the VBE I added) put back the resistors.
To apply each mosfet the Zobel I have to redesign the whole layout
Before I write anything wrong I go back to study the original circuit was built by Cordell himself.
I will mode the layout so the mentioned transistor pair be mounted on the main heatsink , simple solution to extend the transistor leads using some wires.
Thank you
Now I see in real life how to apply that type of Zobel you refer or advise.
But do I have to go there
These circuit was built by Cordell himself with Hitachi mosfets. I assume you know about that.
The difference I want to use these Toshiba 2SK1529 & 2SJ200
The reason for changing the mosfet I have those at least 8 years and I never find a good circuit where I can use them.
OK I go back to the original circuit (remove the VBE I added) put back the resistors.
To apply each mosfet the Zobel I have to redesign the whole layout
Before I write anything wrong I go back to study the original circuit was built by Cordell himself.
I will mode the layout so the mentioned transistor pair be mounted on the main heatsink , simple solution to extend the transistor leads using some wires.
Thank you
Attachments
Let s go back to the designer orig circuit
The output stage
I do not want over complicate these circuit, to be honest I do not have the knowledge to try to improve 😀 these circuit!
I'm nobody compare to Cordell at least my knowledge about audio design under the barrel compare to his.
I just want to make these work with my Toshiba devices by doing only the necessary mode.
If that is not possible in a simple way I will have to give up my dream.
Doing that I will end up with the Hitachi or other lat-fets (I do have them but several amp I built with those devices and I have mixed feeling about them)
metallicus69 here is the original circuit I mentioned my previous posts.
Greetings gabor
The output stage
I do not want over complicate these circuit, to be honest I do not have the knowledge to try to improve 😀 these circuit!
I'm nobody compare to Cordell at least my knowledge about audio design under the barrel compare to his.
I just want to make these work with my Toshiba devices by doing only the necessary mode.
If that is not possible in a simple way I will have to give up my dream.
Doing that I will end up with the Hitachi or other lat-fets (I do have them but several amp I built with those devices and I have mixed feeling about them)
metallicus69 here is the original circuit I mentioned my previous posts.
Greetings gabor
If you want to go the lateral mosfet route then I found ALF16N16W and ALF16P16W to be good lateral's. They are 250 watts so are powerful.
Those Toshiba devices are really nice for vertical fets. It may not be a requirement to use local EC or Hawksford's amplified diode circuit if it seems too complicated, at least starting out. It might be neat to try them as if they were Laterals, with just the addition of a Vbe multiplier. Either way, I will say that it is better to place the gate Zobel between the gate pin and drain pin. Here is a post that pertains to the gate Zobel filter.
If HEC is on your mind then Here is an example of this circuit used, but the devices that are used here are planer stripe mosfets, significantly less linear than those Toshiba devices.
As far as thermal compensation, the photo below shows the error devices (small SMD inside yellow box) mounted just under and in contact with the drain pin of the output devices. This method seems to work really good and yet keeps the circuit layout small. At the end of the thread, the bias is completely off (class B) just to show what difference can be made with EC.
In another thread, Here is an example of the error signal that is generated by the HEC circuit when driving non-linear devices such as planer stripe fets. Compare the On-region characteristics of those planer stripe fets to that of the Toshiba devices.
If HEC is on your mind then Here is an example of this circuit used, but the devices that are used here are planer stripe mosfets, significantly less linear than those Toshiba devices.
As far as thermal compensation, the photo below shows the error devices (small SMD inside yellow box) mounted just under and in contact with the drain pin of the output devices. This method seems to work really good and yet keeps the circuit layout small. At the end of the thread, the bias is completely off (class B) just to show what difference can be made with EC. In another thread, Here is an example of the error signal that is generated by the HEC circuit when driving non-linear devices such as planer stripe fets. Compare the On-region characteristics of those planer stripe fets to that of the Toshiba devices.
Attachments
Today I mod. the layout Q72 & Q71 now will be mounted on the main heatsink!
Thank you for pointing it out, that would cause problem (thermal runaway very likely) even with Lat-fet.
I see Cordell did mounted between the power mosfets.
Do you advise to get read of the VBE because of the HEC may be just use diodes
Or leave it how it was with the resistors and reduce their value.
Right now neither placed on the PC board I do wait on your answer.
If I use Zobel at each power device (Cordell didn't built that way ) the whole layout has to be redesigned.
That case I need the ground trace close to each power mosfet. To much work.
nigelwright7557
I do have high power EXICON (similar you mentioned) latFet but let try to adopt to the Toshiba type.
🙂
Attachments
Hi Gabor,
I guess the way forward is quite clear, even in the link you provided:
We will not give details here for anything else but lateral MOSFETs output devices. Enough to know that adapting for verticals or BJT involves three steps:
1. Strap the PCB for the correct output device pin-out
2. Adjust the R75/R76/R88 voltage divider for the power devices bias (Note 1 below). Fine adjustments can be done via R78
3. Choose the gate/base resistor. Vertical MOSFETs need lower values while BJTs need even lower values (if at all!).
4. Adjust the Hawksword balance (either R71/R72 or R81/R20) for a minimum in THD. On the Hawksword balance adjustment see Note 2 below.
the Toshiba devices have VgsON similar to Laterals, so that part can stay the same (point 2). The Hawksword balance is more tricky (point 4), and requires access to a THD meter. Point 3 is also to be taken into account. You have to decrease the resistor on the gate but not too much, as you will have oscillations. Too bad you cannot implement the Zobel's as those can be left out in case they are not needed.
Thanks for your efforts!
I guess the way forward is quite clear, even in the link you provided:
We will not give details here for anything else but lateral MOSFETs output devices. Enough to know that adapting for verticals or BJT involves three steps:
1. Strap the PCB for the correct output device pin-out
2. Adjust the R75/R76/R88 voltage divider for the power devices bias (Note 1 below). Fine adjustments can be done via R78
3. Choose the gate/base resistor. Vertical MOSFETs need lower values while BJTs need even lower values (if at all!).
4. Adjust the Hawksword balance (either R71/R72 or R81/R20) for a minimum in THD. On the Hawksword balance adjustment see Note 2 below.
the Toshiba devices have VgsON similar to Laterals, so that part can stay the same (point 2). The Hawksword balance is more tricky (point 4), and requires access to a THD meter. Point 3 is also to be taken into account. You have to decrease the resistor on the gate but not too much, as you will have oscillations. Too bad you cannot implement the Zobel's as those can be left out in case they are not needed.
Thanks for your efforts!
Thank you !!!
1) it is done
2) these can be done after some measurement
3) I did lower the gate resistors from 470R to 220R, in case you think better to lower even more let say 100R I'm not against that
4) I will do that
If I run into oscillation that will be a big problem (I would need a scope at least)
If it happen I will have to back to the original design Hitachi mosfet😱
at Zobel the C8 & C9 value are 39pF ? May be I will adopt the layout
Thank you one more time
Attachments
Thanks for your input! If Q4 Q8 are to be used also for thermal management/monitoring for the outputs, how is this done with E_Line/TO-92 devices (ZTX653/753)?
The trick is reconnection of emitter resistors. So correction current dont disturb the emitters of 1-st stage.
I dont so English speaker to understand your humor🙂 If you mean "original" which in your schematic - use it. DC points do not be affected too much Trimmers will help.
- Status
- Not open for further replies.