Hi
I try to build Mike Renardson´s MJR7-mk5 amplifier. I made my own board design but someting is wrong. I have now tied to find the problem, butt no sucscess.
The 2A fuse bloves when i power on, but not a 2.5A fuse. With a 2.5A fuse the R13 gets extremely hot. If I messure without T6 and T7 connected I get 40V at every point around Vr1. The rail are 67V.
I staring me blind at this so I can´t se the problem. Can someone check my drawings?
I try to build Mike Renardson´s MJR7-mk5 amplifier. I made my own board design but someting is wrong. I have now tied to find the problem, butt no sucscess.
The 2A fuse bloves when i power on, but not a 2.5A fuse. With a 2.5A fuse the R13 gets extremely hot. If I messure without T6 and T7 connected I get 40V at every point around Vr1. The rail are 67V.
I staring me blind at this so I can´t se the problem. Can someone check my drawings?
The gate and drain on T6 are swapped in the schematic but appear ok on the board. Perhaps a diode is wrong way around? R13 getting hot would indicate a large current flow through there with the only paths being the gate diodes or T7 itself. With the outputs removed there is no dc feedback, so I wouldn't expect the voltages to be right at all. Not really a fan of how dc offset is controlled. Looks like it would vary a fair amount with load or no load.
During testing or powering on be sure you have connected a dummy load to get the offset right...I did this amp several years ago and encountered problems like cold solder and no load testing..When i found the cold solder and put the load it works fine and use the amp till now.
At no load offset is high...
At no load offset is high...
Hi Livius
I have looked at your diagrams but don't see anything seriously wrong. There is one error, you have used a polarised capacitor for C13, which was specified in the parts list as a non-polar electrolytic. If you don't have one available the polarised type may be ok but needs to be reversed, with the positive side going to the junction of R8 and R14. Also on the board layout VR2 appears to have all 3 pins connected together.
My first guess would be a faulty mosfet, a shorted drain to source for T6 could explain the blown fuse and high current through R13, but I believe these mosfets are more likely to fail open rather than short-circuit, so maybe also check for a short on the board. Measuring the drain-source and gate-source voltages for T6 and T7 may give some idea what is wrong.
Regards, Mike
I have looked at your diagrams but don't see anything seriously wrong. There is one error, you have used a polarised capacitor for C13, which was specified in the parts list as a non-polar electrolytic. If you don't have one available the polarised type may be ok but needs to be reversed, with the positive side going to the junction of R8 and R14. Also on the board layout VR2 appears to have all 3 pins connected together.
My first guess would be a faulty mosfet, a shorted drain to source for T6 could explain the blown fuse and high current through R13, but I believe these mosfets are more likely to fail open rather than short-circuit, so maybe also check for a short on the board. Measuring the drain-source and gate-source voltages for T6 and T7 may give some idea what is wrong.
Regards, Mike
Thank´s for the replies.
I mounted a non polar cap on the bord, but I did not find one in Eagle. Thats why ther is a polar cap on the drawing.
Replacing T4 and T5 to new ones did not help.
I meassured T6 and T7 with a multimeter and diode treshold voltage measurement between D and S. T6 gave aprox 0.6 V, and T7 aprox 0.3V. I will try to replace T7. I have two more of the L-mostets.
I will check the made board again for shorts.
I will also use 6 ohm resistor on the output.
Strange, the board i made do not have shorted Vr2. I vill check it in the Eaglefile.
I mounted a non polar cap on the bord, but I did not find one in Eagle. Thats why ther is a polar cap on the drawing.
Replacing T4 and T5 to new ones did not help.
I meassured T6 and T7 with a multimeter and diode treshold voltage measurement between D and S. T6 gave aprox 0.6 V, and T7 aprox 0.3V. I will try to replace T7. I have two more of the L-mostets.
I will check the made board again for shorts.
I will also use 6 ohm resistor on the output.
Strange, the board i made do not have shorted Vr2. I vill check it in the Eaglefile.
Hi junior35
Sorry, I don't have any boards now. I made some a few years ago, but all were sold. Now I don't even have the equipment to make them. The last time I made something for my own use I used an etch-resist pen to make the board.
If you look in the Vendor's Bazaar in the commercial sector here on diyAudio there is someone called msdesigner who makes pc boards, and I think he can make boards from image files, so that may be worth trying.
Mike
Sorry, I don't have any boards now. I made some a few years ago, but all were sold. Now I don't even have the equipment to make them. The last time I made something for my own use I used an etch-resist pen to make the board.
If you look in the Vendor's Bazaar in the commercial sector here on diyAudio there is someone called msdesigner who makes pc boards, and I think he can make boards from image files, so that may be worth trying.
Mike
Hi junior35
There was one small addition I suggested to improve reliability. The 220R preset is used to set quiescent current, and it is said that these controls can fail with an open-circuit wiper. I have never had one of these fail myself, but some designers say the control should be connected in such a way that the quiescent current falls rather than increases with an open-circuit, which seems reasonable. With my circuit it is not entirely certain what would happen. Can these mosfets have any leakage current from drain to gate? That could make the gate voltages rise. I would guess this is not a problem, but don't know for certain, so I suggest adding something like 5k6 in parallel with the 0.1u capacitor connected to the control. That doesn't worry me enough to add the resistor to my own amplifiers.
The only other problem I know about was mentioned earlier by jerluwoo. The 4k7 preset is used to set the mosfet source voltage to half the supply voltage, but this requires a speaker or low resistance load to be connected while making the adjustment. There are alternative arrangements which could avoid this, for example connecting the 4k7 preset from the 56k resistor to 0V, but that causes a different problem with added low frequency phase shift in the feedback loop and a danger of instability depending on source and load impedance. Having the 4k7 in parallel with the 10u feedback capacitor only causes a small problem during setup, not in normal use, so I am still happy with that.
Mike
There was one small addition I suggested to improve reliability. The 220R preset is used to set quiescent current, and it is said that these controls can fail with an open-circuit wiper. I have never had one of these fail myself, but some designers say the control should be connected in such a way that the quiescent current falls rather than increases with an open-circuit, which seems reasonable. With my circuit it is not entirely certain what would happen. Can these mosfets have any leakage current from drain to gate? That could make the gate voltages rise. I would guess this is not a problem, but don't know for certain, so I suggest adding something like 5k6 in parallel with the 0.1u capacitor connected to the control. That doesn't worry me enough to add the resistor to my own amplifiers.
The only other problem I know about was mentioned earlier by jerluwoo. The 4k7 preset is used to set the mosfet source voltage to half the supply voltage, but this requires a speaker or low resistance load to be connected while making the adjustment. There are alternative arrangements which could avoid this, for example connecting the 4k7 preset from the 56k resistor to 0V, but that causes a different problem with added low frequency phase shift in the feedback loop and a danger of instability depending on source and load impedance. Having the 4k7 in parallel with the 10u feedback capacitor only causes a small problem during setup, not in normal use, so I am still happy with that.
Mike
Hi and Happy New Year.
No progress here with the amplifier.
I have tried all my mosfets with the same result. So I have ordered new 2SK1058 and 2SJ162, from a suplier in GB (Cricklewood electr.). Those I have are from a Chinese supplier. Fake?
So now I am waiting........
If the new mosfets are working then I will make new pcb´s. I will ad the posibility to ad a resistor paralell to C10.
No progress here with the amplifier.
I have tried all my mosfets with the same result. So I have ordered new 2SK1058 and 2SJ162, from a suplier in GB (Cricklewood electr.). Those I have are from a Chinese supplier. Fake?
So now I am waiting........
If the new mosfets are working then I will make new pcb´s. I will ad the posibility to ad a resistor paralell to C10.
Hi
I found one pair of 2SJ50 and one pair of 2SK135 in one of my boxes. I tried these with the amplifier. Guess what. It worked.
That means the mosfets from china are fake. So be careful bying from china.
Is it possible to use J50 and K135 in this construction? The same data as J162 and K1058.
I found one pair of 2SJ50 and one pair of 2SK135 in one of my boxes. I tried these with the amplifier. Guess what. It worked.
That means the mosfets from china are fake. So be careful bying from china.
Is it possible to use J50 and K135 in this construction? The same data as J162 and K1058.
I had experience of fake devices like k1057/k1058/J161/j162. When i tested them they are Vertical fets and pins are different..instead of GSD for laterals they actually GDS.
I had increased my feedback capacitor from 10 PF to 22Pf and oscillation is gone..maybe caused by board and wiring layout and change of transistors..
I had increased my feedback capacitor from 10 PF to 22Pf and oscillation is gone..maybe caused by board and wiring layout and change of transistors..
Hi Livius
That's good to hear, you found the problem. The J50 and K135 should work ok in this circuit, they are the same specification apart from the TO3 case, so if you have a suitable heatsink and insulators there is no reason not to use them. As mentioned by Junm some fakes are just re-marked vertical mosfets with the centre pin connected to the drain. If both the mosfets you used were verticals that should just blow the fuse, but your observation that a higher rated fuse survived and R13 got hot suggests maybe only the k1058 is a fake vertical and the J162 could be a genuine lateral. Anyway, I'm happy to hear you now have a working amplifier.
Mike
That's good to hear, you found the problem. The J50 and K135 should work ok in this circuit, they are the same specification apart from the TO3 case, so if you have a suitable heatsink and insulators there is no reason not to use them. As mentioned by Junm some fakes are just re-marked vertical mosfets with the centre pin connected to the drain. If both the mosfets you used were verticals that should just blow the fuse, but your observation that a higher rated fuse survived and R13 got hot suggests maybe only the k1058 is a fake vertical and the J162 could be a genuine lateral. Anyway, I'm happy to hear you now have a working amplifier.
Mike
On second thoughts probably both fakes are n-channel vertical mosfets. If someone was making fakes I don't think they would care whether they were n or p-channel, they would use the same type for both fakes. If the 'J162' was a n-channel vertical instead of a p-channel lateral it would now have the correct supply polarity from drain to source and the observed high current with hot R13 is then possible.
Mike
Mike
Verticals would be hard to implement without some serious redesign i would have thought - much higher Vgs plus the necessary temperature compensation, higher gate capacitances, compensation changes.
If you cant find the specified fets, the laterals from Semelab are a good (and available!) substitute. They are in stock at Farnell for example. http://sg.element14.com/webapp/wcs/...tegoryId=800000029007&langId=65&storeId=10191
Separately, i have always admired the elegance of this design - and can't help but wonder how much more popular this amp would be if the output capacitor were dispensed with.
If you cant find the specified fets, the laterals from Semelab are a good (and available!) substitute. They are in stock at Farnell for example. http://sg.element14.com/webapp/wcs/...tegoryId=800000029007&langId=65&storeId=10191
Separately, i have always admired the elegance of this design - and can't help but wonder how much more popular this amp would be if the output capacitor were dispensed with.
Last edited:
I suspect that the same guys who would find a dual power supply and no output cap. more attractive, would also like to see an LTP input stage, Hexfets and so on until the design was just another ho-hum generic design that then needed protection circuits, thermal compensation of bias, output relays and generally more complexity.
Profusion plc also supply the same complementary pairs as Semelab's ALFET types in their own Exicon house brand. i.e: ECX10N20/10P20 in lieu of ALF08N16V/08P16V
Hi junior35
I agree with the replies from kasey197 and Ian Finch, I think they have covered everything. As for output capacitors, it is an easy solution to speaker protection, and including the capacitor inside the main feedback loop avoids any noticeable effect on distortion or damping factor, but I accept that it will reduce the popularity of the design. The lateral mosfets easily available here in UK include the Semelab ALF08P16V and ALF08P16V, but on the UK-Farnell site they are listed as 'Available until stock is exhausted' and 'No longer manufactured', so availability in the future may become more of a problem.
Mike
I agree with the replies from kasey197 and Ian Finch, I think they have covered everything. As for output capacitors, it is an easy solution to speaker protection, and including the capacitor inside the main feedback loop avoids any noticeable effect on distortion or damping factor, but I accept that it will reduce the popularity of the design. The lateral mosfets easily available here in UK include the Semelab ALF08P16V and ALF08P16V, but on the UK-Farnell site they are listed as 'Available until stock is exhausted' and 'No longer manufactured', so availability in the future may become more of a problem.
Mike
Hi
I have got my new J162 and K1058. And thees ones works well.
I also tried J50 and K135. There are a little differece between the old and the new types. With J162 and K1058 I can't set quiescent qurrent higher then 185 mA, but with the J50 and K135 it's possible to get aprox 300 mA. It's a stable design, even when the temperature gets high.
I have no good test equipment (a cheap DSO to my computer) for testing SNR and distortion. So I can't say anything about that.
I tried this first plate with a cheap speaker and I did not know that this speaker can give such a "good" sound. I will build one more and test it with my ordinary HiFi equipment.
I have got my new J162 and K1058. And thees ones works well.
I also tried J50 and K135. There are a little differece between the old and the new types. With J162 and K1058 I can't set quiescent qurrent higher then 185 mA, but with the J50 and K135 it's possible to get aprox 300 mA. It's a stable design, even when the temperature gets high.
I have no good test equipment (a cheap DSO to my computer) for testing SNR and distortion. So I can't say anything about that.
I tried this first plate with a cheap speaker and I did not know that this speaker can give such a "good" sound. I will build one more and test it with my ordinary HiFi equipment.
Lateral mosFETs are like all other FETs in that production spread of parameters can be enormous.
The latFETs show a range of Vgs(off) from 0.15V to 1.5V
that's 1:10, as I said "enormous". This determines the voltage you need to apply to turn on the FET.
The typical gm is ~1A/V. i.e. increase the Vgs by 1mV and the current increases by ~1mA. This too varies, Yfs from 0.7A/V to 2A/V
If you already have 185mA and you want 300mA, then you need an INCREASE of Vgs of ~(300-185) = an extra ~115mVgs
But you have upper and lower devices. You need an extra 230mV of bias voltage.
A higher resistor value will give you a higher bias voltage. Try the next higher E12 value and see what it adjusts your current to.
The latFETs show a range of Vgs(off) from 0.15V to 1.5V
that's 1:10, as I said "enormous". This determines the voltage you need to apply to turn on the FET.
The typical gm is ~1A/V. i.e. increase the Vgs by 1mV and the current increases by ~1mA. This too varies, Yfs from 0.7A/V to 2A/V
If you already have 185mA and you want 300mA, then you need an INCREASE of Vgs of ~(300-185) = an extra ~115mVgs
But you have upper and lower devices. You need an extra 230mV of bias voltage.
A higher resistor value will give you a higher bias voltage. Try the next higher E12 value and see what it adjusts your current to.
Last edited: