hackernap advice thread
- Thread starter Lucaslucas
- Start date
keithsnaim
Gone Fishing
Thanks Carl .
You could swap out R30 and R34 for values of 1.5x or 2x the value. So for example, if you fitted 1M resistors, just fit 1.5M or even 2M. What this does is reduce the voltage drop across the VBE from 8V to 6V or 4V.
There are no side effects to doing this, although I wouldn't go much higher than 2M Ohms because you ideally want at least a 4V drop across the VBE.
So try changing R30 and R34 from 1M Ohms to 1.5M Ohms and measure the front end voltages again. You should find that they're just a tad higher than your output voltage, all being well
Hi and sorry for the resurrection of this old thread but I need to ask some questions about the VBE, please!
Although I use 2MOhm as R30 & 34 my VBE circuit is dropping 6.5V instead of 4V and I cannot reach my OP 53V voltage
Can I experiment with even higher than 2M values in order to reduce the voltage drop or this is not a good solution? Shall I look for some other problems inside my build because of this higher than normal voltage drop, please?
Thank you all!
Ivo
john.luckins
pfm Member
Hi Ivo,
The Hackernap operates the IRF 610/9610's at the bottom end of their current range and they also can have a wide variation in Vds for a given load current. It is off the useable scale on the datasheet graphs. There's a couple of things worth doing before you change the values of R28/R33 and/or R30/R34. Firstly very carefully measure the voltage you get at the junction of R28 & R30 and similarly R33 & R34 and chech that these voltages are what you might expect the voltage divider to produce. With a 57 volt input you should get around 54.3 volts. Secondly check the currents going into the fron tend Vbe to ensure it is around 14mA per channel +ve and 33mA -ve. If it is significantly higher then there are other issues causing the high Vds.
If all is well then you can change the values of the voltage divider resistors, but I would be wary of going much higher than 2M for R30 & R34 because its value will become significant with respect to the leakage current in C16/C18. I would suggest that you drop the value of R28/R33 to say 91k or 82k instead first.
John
The Hackernap operates the IRF 610/9610's at the bottom end of their current range and they also can have a wide variation in Vds for a given load current. It is off the useable scale on the datasheet graphs. There's a couple of things worth doing before you change the values of R28/R33 and/or R30/R34. Firstly very carefully measure the voltage you get at the junction of R28 & R30 and similarly R33 & R34 and chech that these voltages are what you might expect the voltage divider to produce. With a 57 volt input you should get around 54.3 volts. Secondly check the currents going into the fron tend Vbe to ensure it is around 14mA per channel +ve and 33mA -ve. If it is significantly higher then there are other issues causing the high Vds.
If all is well then you can change the values of the voltage divider resistors, but I would be wary of going much higher than 2M for R30 & R34 because its value will become significant with respect to the leakage current in C16/C18. I would suggest that you drop the value of R28/R33 to say 91k or 82k instead first.
John
Hi John,
Thanks for the informative (as usual) reply!
Unfortunately, I left my amp boards at the office and will report my measurements on Monday. The strange thing is that both channels are acting the same i.e. 57.5V input and around 51 V after the VBE circuit - hope I didn't do something wrong!
Thanks again!
Ivo
Thanks for the informative (as usual) reply!
Unfortunately, I left my amp boards at the office and will report my measurements on Monday. The strange thing is that both channels are acting the same i.e. 57.5V input and around 51 V after the VBE circuit - hope I didn't do something wrong!
Thanks again!
Ivo
OK, so I did some measurements on one of the boards this morning and the results are:
FE+ = 57.9V, VBE+ out = 51.5V
FE- = 57.8V, VBE- out = 51.1V
V R28/30 = 54.3V
V R33/34 = 54.1V
I FE+ = 11.4mA
I FE- = 13.7mA
I suppose, everything looks normal but really don't know why is this higher VBE voltage drop of 6.5V with R30(34) = 2MOhms?
I will try John's suggested combo of 82K/2M for R28/R30 but also saw Teddy's recommendation for 64K/1.5M - which one is better, please?
Thank you!
Ivo
FE+ = 57.9V, VBE+ out = 51.5V
FE- = 57.8V, VBE- out = 51.1V
V R28/30 = 54.3V
V R33/34 = 54.1V
I FE+ = 11.4mA
I FE- = 13.7mA
I suppose, everything looks normal but really don't know why is this higher VBE voltage drop of 6.5V with R30(34) = 2MOhms?
I will try John's suggested combo of 82K/2M for R28/R30 but also saw Teddy's recommendation for 64K/1.5M - which one is better, please?
Thank you!
Ivo
john.luckins
pfm Member
It is a compromise. If you drop the values of R28 and R33 too much you lose some of the lowest frequency filtering. If you raise R30 and R34 too high then the circuit becomes more sensitive to cap leakage currents and noise. I doubt whether I could hear the difference between the two let alone judge which was best, but Teddy has spent more time on this. We're in the right ballpark.
The voltages you are getting off the resistive dividers seem too low and to indicate that R28 and R33 are closer to 130K or that there are leakage currents through C16 or C17. The current through the 2M resistors is only around 27uA so if the upper resistors are 100k then there seems to be about 9uA of leakage. What sort of caps are you using for C16 and C17?
If you accept the leakage current, which is probably OK to do because it is filtered, keep with 2Megg for R30/34 and reduce R28 and R33 to 75 ohms you will get roughly a volt less across the FET than you are now. Track down and remove the leakage current and you'll get that volt back with no change to the resistors.
john
The voltages you are getting off the resistive dividers seem too low and to indicate that R28 and R33 are closer to 130K or that there are leakage currents through C16 or C17. The current through the 2M resistors is only around 27uA so if the upper resistors are 100k then there seems to be about 9uA of leakage. What sort of caps are you using for C16 and C17?
If you accept the leakage current, which is probably OK to do because it is filtered, keep with 2Megg for R30/34 and reduce R28 and R33 to 75 ohms you will get roughly a volt less across the FET than you are now. Track down and remove the leakage current and you'll get that volt back with no change to the resistors.
john
John, thanks!
C16(18) are 10uF,63V Evox MMK and C17(19) are 100nF Evox SMR - I suppose they are good?
For C5(8) though I am using some old 10uF Siemens film caps I found here - could they be the problem, please?
I will check if R28,29,32,33 are correct 100K values, as well.
Ivo
C16(18) are 10uF,63V Evox MMK and C17(19) are 100nF Evox SMR - I suppose they are good?
For C5(8) though I am using some old 10uF Siemens film caps I found here - could they be the problem, please?
I will check if R28,29,32,33 are correct 100K values, as well.
Ivo
john.luckins
pfm Member
Hi Ivo,
The caps within the resistive divider chain are low leakage types so I'm puzzled. Of course when you are measuring voltages across high resistances the input resistance of the meter becomes important too and could register as a lower voltage reading across the 2meg resistors. Depends upon the meter. If you had a scope the measurement might be more accurate. Either way I would discount cap leakage (any C5 leakage wouldn't cause any trouble in the resistive divider so that's OK). I suspect that lost 1 volt is actually part of the Vgs figure when you aren't measuring anything. These fets have quite a wide range of Vgs for a given drain current and it varies a lot with temperature. I'd try 75k in for R28 and R30 and see if you gain a volt at the output. An easy way to check this without desoldering anything is just to tack a resistor across R28 and R30. If you use the old 1meg resistor the resulting combination will give you approx. 70k which is close enough to prove the point.
Good idea on checking all the resistors and perhaps give the board a clean around the junction in the divider to remove any residual flux. It is a sensitive node.
John
PS, if you measure the voltage across R28 and R30 any meter leakage will be less significant and will give you a more accurate figure for the true resistive divider voltage.
The caps within the resistive divider chain are low leakage types so I'm puzzled. Of course when you are measuring voltages across high resistances the input resistance of the meter becomes important too and could register as a lower voltage reading across the 2meg resistors. Depends upon the meter. If you had a scope the measurement might be more accurate. Either way I would discount cap leakage (any C5 leakage wouldn't cause any trouble in the resistive divider so that's OK). I suspect that lost 1 volt is actually part of the Vgs figure when you aren't measuring anything. These fets have quite a wide range of Vgs for a given drain current and it varies a lot with temperature. I'd try 75k in for R28 and R30 and see if you gain a volt at the output. An easy way to check this without desoldering anything is just to tack a resistor across R28 and R30. If you use the old 1meg resistor the resulting combination will give you approx. 70k which is close enough to prove the point.
Good idea on checking all the resistors and perhaps give the board a clean around the junction in the divider to remove any residual flux. It is a sensitive node.
John
PS, if you measure the voltage across R28 and R30 any meter leakage will be less significant and will give you a more accurate figure for the true resistive divider voltage.
Thank you, John!
Unfortunately some work appeared at the office and I have to postpone again the whole project for a week or two.
I will do the test you suggested but first will try to re-do all the soldier joints around the VBE - they are not that much anyway.
I will report back my results soon.
Thanks,
Ivo
Unfortunately some work appeared at the office and I have to postpone again the whole project for a week or two.
I will do the test you suggested but first will try to re-do all the soldier joints around the VBE - they are not that much anyway.
I will report back my results soon.
Thanks,
Ivo
I haven't looked extensively yet, but can't find the following parts from the BOM at either Mouser, Digikey or Farnell
C4 - 39pF Polystyrene film cap (preferred) OR ...
C15, C17, C19 - 100nF polypropylene film cap
C24 - 220nF polypropylene film cap
Also, R21, R22 and R23 are out of stock at Mouser with an 11 week lead time, as is R36 with a 10 week lead time.
I find the Digikey product search bloody useless. I want to shift some of my Mouser order over to Digikey to get free delivery on a £50 order with both. Any suggestions on how to use the component search effectively would be most welcome. Or any suggestions on DIY electronics products that Digikey do at competitive prices, so I can bump up my order that way, would also help.
C4 - 39pF Polystyrene film cap (preferred) OR ...
C15, C17, C19 - 100nF polypropylene film cap
C24 - 220nF polypropylene film cap
Also, R21, R22 and R23 are out of stock at Mouser with an 11 week lead time, as is R36 with a 10 week lead time.
I find the Digikey product search bloody useless. I want to shift some of my Mouser order over to Digikey to get free delivery on a £50 order with both. Any suggestions on how to use the component search effectively would be most welcome. Or any suggestions on DIY electronics products that Digikey do at competitive prices, so I can bump up my order that way, would also help.
There are a few inconsistencies in the BOM v1.7
R4 : Type = 2K7, Schematic = 2K7, Mouser Link = 3K
R13 : Type = 27K, Schematic = 27K, Mouser Link = 26K7
R30, R34 : Type = 2M, Schematic = 1M, Mouser Link = 1M
I've marked what I assume to be the correct values in bold. Could someone please confirm?
R4 : Type = 2K7, Schematic = 2K7, Mouser Link = 3K
R13 : Type = 27K, Schematic = 27K, Mouser Link = 26K7
R30, R34 : Type = 2M, Schematic = 1M, Mouser Link = 1M
I've marked what I assume to be the correct values in bold. Could someone please confirm?
orangeart
KJF Audio Ltd.
The schematic for the original naim circuit and the ncc200 with is where this comes from in the first place are 2k7. I seem to remember back in the BOM 1.0 days that there wasn't a 2k7 available and the GB that happened used the 3k hence the link. I think the same happened for the R13
If you are using transformers of the specced voltage it was decided that the 2M resistor was better, obviously the schematic and link were not updated.
Stefan
If you are using transformers of the specced voltage it was decided that the 2M resistor was better, obviously the schematic and link were not updated.
Stefan
337alant
Negatively Biased
I follow your logic for R13, however for R4 the same logic doesn't work for me. If it's fine to use 2K7 or 3K and 26K7 or 27K, I can't see the point in expensive devices with 0.1% tolerance. I'll take a peek at the original naim/ncc200 circuit.
For R30, R34 I've got the spec'ed traffos so I'll use 2M. Thanks Stefan.
For R30, R34 I've got the spec'ed traffos so I'll use 2M. Thanks Stefan.
But what resistance value(s)?
James Evans
Bedroom Bodger
I believe the tight tolerance is specified for channel matching, rather than absolute schematic accuracy.