The P-side and N-side devices have different balances by their very fabrication. It is normal and expected that you would have different bias on the top and the bottom halves to maintain balance between the two currents and steady the output to zero.
If you face an issue of severe imbalance, you might try JFETs that are more dissimilar (Patrick/EUVL has a nice read about it). This means selecting something like a GR grade J74 to go with a BL K170, for example. The P-channel devices have higher transconductance, and often matched JFETs aren't the right way - you need a set, not a match. I think from memory the difference was about 2mA for optimum 'match' in device transconductance.
The later addition of P3 was meant to solve this to some extent, and you can try moving it to better realise a closer bias match for zero offset. This is not the correct or recommended way to tune P3, but it can still be used for this purpose. You should stop thinking about voltages (I find mV bias readings almost impossible to troubleshoot with) and start thinking about currents. Mismatched currents could also be caused by mismatched resistor tolerances, or resistors at the end of their tolerance spectra (5% resistors can be 10% apart, for example, and still meet spec). Sometimes effects will cascade and add negatively, causing small differences to sum to a whopper.
You can measure the current through the power supply series resistors to get a baseline current value for the upper and lower halves of the circuit. What matters is what the power supply is seeing. Even there you can expect some differences, but it should be fairly small.
If you face an issue of severe imbalance, you might try JFETs that are more dissimilar (Patrick/EUVL has a nice read about it). This means selecting something like a GR grade J74 to go with a BL K170, for example. The P-channel devices have higher transconductance, and often matched JFETs aren't the right way - you need a set, not a match. I think from memory the difference was about 2mA for optimum 'match' in device transconductance.
The later addition of P3 was meant to solve this to some extent, and you can try moving it to better realise a closer bias match for zero offset. This is not the correct or recommended way to tune P3, but it can still be used for this purpose. You should stop thinking about voltages (I find mV bias readings almost impossible to troubleshoot with) and start thinking about currents. Mismatched currents could also be caused by mismatched resistor tolerances, or resistors at the end of their tolerance spectra (5% resistors can be 10% apart, for example, and still meet spec). Sometimes effects will cascade and add negatively, causing small differences to sum to a whopper.
You can measure the current through the power supply series resistors to get a baseline current value for the upper and lower halves of the circuit. What matters is what the power supply is seeing. Even there you can expect some differences, but it should be fairly small.
I have 400VA 30V secondary and I get somewhere between 36-39V DC with load. Just check other posts on forum so that you don't fall short of target.
I would suggest bigger the heatsinks the better, you always have the option to bias higher and mosfets will remain bit cooler.
I was at 350mV and after adding another pair of JFET I am at 500mV, just repositioning the mosfets, will go higher and see what is the difference.
I would suggest bigger the heatsinks the better, you always have the option to bias higher and mosfets will remain bit cooler.
I was at 350mV and after adding another pair of JFET I am at 500mV, just repositioning the mosfets, will go higher and see what is the difference.
Thanks for the replies. I’m thinking 350mv bias. I’ve built a few different projects using the normal full size modushop chassis, one of them being the F5T V2 as a dual monaural stereo amp in the 5U Deluxe and it’s fine. I want to go separate mono-blocks this time around and would like something that looks a bit different but I don’t want to run into heat issues either.
Last edited:
No one has chimed in... I'll try.
What does the article state is the dissipation for a 'normal' F5TV3?
What dissipation did you calculate for your build at the bias and rail voltages you're considering?
How does the dissipation you're proposing mesh against what others have said they've experienced (and/or Modushop's posted data) for temps at that dissipation on those heatsinks?
What does the article state is the dissipation for a 'normal' F5TV3?
What dissipation did you calculate for your build at the bias and rail voltages you're considering?
How does the dissipation you're proposing mesh against what others have said they've experienced (and/or Modushop's posted data) for temps at that dissipation on those heatsinks?
Thanks for chiming in. To answer a couple of your questions, first of all let me state, I don’t know how to calculate my heat dissipation, thus is why I came to the group to ask for help. Secondly, I’ve read the article numerous times and although I’m sure it’s stated and those who may understand how to decipher it, I unfortunately don’t. I have looked over 6l6’s build guide several times and also see where he states “The 5U and 4U Deluxe chassis from the DIY audio store are basically identical except for size and that the 5U heatsinks are 2 pieces per side”. The Modushop mono case from I understand is a 4U style chassis and can be fitted with sinks on both sides. What I’m trying to avoid is dropping an excess of $750 on a couple chassis’s if they’re not going to do what I need them to do. But once again, I don’t know how to calculate this info. Lastly, im not sure if anyone else is experiencing any issues but on my end, the “New and improved” DIY Store information about the chassis is very limited and it looks like there’s only limited items stocked in the states. Going directly to the modushop website is, at least on my tablet, limited as well. When I open pages to get data, I get errored pages. Thus reaching out to the group.
Check out this thread, some useful info on the Store heatsinks: https://www.diyaudio.com/community/threads/the-sit-3x-amplifier.353999/page-37#post-6611333
Rush
Rush
Totally understand the motivation... so... I'll try to help. I'll also walk through how I might answer my own questions b/c I'm often wrong, but people are typically kind enough to correct me. 
A quick search in the article for "dissipation" and clicking a few times... shows just after "Here is the V3".... "The differences are found in the doubling of the number of parallel output devices and their associated parts for 240 watts of dissipation and the addition of Q7 and Q8 and the networks that bias them."
So, using your excellent search skills, you can likely find more information related to the other questions from the article, most important being the rail voltages and bias current that leads to the 240 Watts of dissipation quoted.
Total dissipation is the dissipation of each device all added up. Dissipation for each device is the power through each device. VxA. V is derived from the rail voltages, and A is the current through the two source resistors for each device. You stated a "bias voltage" you want. Translate that to the bias current using Ohm's law and calculate the dissipation for each device. Then... add up the total number of devices. Note - if you ever plan to push it into Class A/B ... it will be higher. That can come later.
That's how I'd start to tackle the problem along with the excellent information from Rush above.
A quick search in the article for "dissipation" and clicking a few times... shows just after "Here is the V3".... "The differences are found in the doubling of the number of parallel output devices and their associated parts for 240 watts of dissipation and the addition of Q7 and Q8 and the networks that bias them."
So, using your excellent search skills, you can likely find more information related to the other questions from the article, most important being the rail voltages and bias current that leads to the 240 Watts of dissipation quoted.
Total dissipation is the dissipation of each device all added up. Dissipation for each device is the power through each device. VxA. V is derived from the rail voltages, and A is the current through the two source resistors for each device. You stated a "bias voltage" you want. Translate that to the bias current using Ohm's law and calculate the dissipation for each device. Then... add up the total number of devices. Note - if you ever plan to push it into Class A/B ... it will be higher. That can come later.
That's how I'd start to tackle the problem along with the excellent information from Rush above.
Last edited:
hello , i found reason . i used 47,5ohm resistor instead of 47.5Kohm for R2. now i have 330 mV on N-side and 270 mv on P-side . that is very similar to second channel i made before.
1. I built an F5T v3 using diyaudio store pcb. V3 can reach 100W. Somewhere on this forum I read that the rail voltage should not go below 45V DC under load. I have 40V AC secondary which gives me 56V DC (No load) but it come down to 50V DC when loaded.
2. Bias depends on, how much heat can you dissipate without burning the devices, otherwise higher bias is good.
Emphasis added in your quote is mine. 1 and 2 are dictated by the fact that you chose 100W and balanced. There is only one answer for output voltage for a certain load. Calculate what you need. See chart below. Many people assume 8 ohm load. If you know your speakers are not 8 ohm, and you want to adjust for 100W at your speaker's highest, lowest, or nominal impedance, go ahead. It's DIY.
Remember that you have double the voltage swing with balanced vs. SE.
I usually assume that I'll lose ~4V from when calculating the rails from desired output voltage.
The current at maximum output power into the load (speaker impedance) you've chosen uses the same math. Iq is dictated by whether you want all of that power in Class A or not. For 100W balanced monoblocks in a typical 5U, I would not set my Iq based on the highest heatsink / device temperatures I could safely achieve, but that's just me.
The other thing you could do for fun is check out any of the First Watt amplifiers that show roughly 100W power output balanced. Many of the First Watt amplifiers use a common transformer / rail voltage. It may help verify your math.
Ask if you need more help.
3 - No.