Hi hall. I finished my f5 turbo v2 with cascode, according to the guide assembly 6l6 and working properly. I've built in two separate monologues with large heatsinks.I'm bias adjustment phase and so far, I have adjusted to 390 mV. Transistors are at 59 degrees, the diodes 41 degrees and 41 degrees heat sinks. The power supply is at 33V.
After searching for the thread it is not clear which is the top of bias. I read that depends on the diodes along with temperature and it seems that sinks can still hold more dissipation, then how have to adjust to take my amplifier to the maximum power?
This amplifier 50w think it is but according to the formula and the maximum recommended bias (400mV) can only reach 40 w for 8 ohm speakers. I would like to reach those 50w How much power develops for 4 ohm speakers? How is class A?
On the other hand, in one of the channels I have 20 mV difference between the N and P, it Is this normal ?, on the other the difference is a couple of mv. I used all matched components. Furthermore, this difference is not whether the bias is set to maximum mv of N and P, the minimum or the average of both.
Excuse my English and my rookie questions, every day I learn !.
Thanks for the forum and provide an opportunity to self-construct audio devices dreams !.
After searching for the thread it is not clear which is the top of bias. I read that depends on the diodes along with temperature and it seems that sinks can still hold more dissipation, then how have to adjust to take my amplifier to the maximum power?
This amplifier 50w think it is but according to the formula and the maximum recommended bias (400mV) can only reach 40 w for 8 ohm speakers. I would like to reach those 50w How much power develops for 4 ohm speakers? How is class A?
On the other hand, in one of the channels I have 20 mV difference between the N and P, it Is this normal ?, on the other the difference is a couple of mv. I used all matched components. Furthermore, this difference is not whether the bias is set to maximum mv of N and P, the minimum or the average of both.
Excuse my English and my rookie questions, every day I learn !.
Thanks for the forum and provide an opportunity to self-construct audio devices dreams !.
Hi Alberto
Your English is fine.
There are basically three limits to the highest bias you can achieve: The power supply capability, the heatsink's dissipation ability and the SOA of the output devices.
With the F5T built according to the guide, the fourth limit is the point where the diode conducts, which is around 400mV. The diodes limit the voltage across the source resistor (or bypass the source resistor, if you choose to look at it another way) to 400mV, hence with them in circuit you will never be able to achieve a reading higher than that even though the amplifier can be running at a higher bias.
In general the safe envelope for each device is around 40W continuous dissipation, though they will be happier at lower operating points. It is also unwise to leave the diodes conducting at zero output because of bias runaway - so if you truly want a higher bias you may have to use lower values of source resistor, so that the conduction point occurs at higher currents. I prefer to add more pairs and run without diodes, but that's probably just me.
And you should watch the power supply voltage, you definitely don't want any sag. Heatsink temperatures look to be under control but 20 degrees difference between the devices and heatsink is an issue though, poor thermal contact maybe?
Your English is fine.
There are basically three limits to the highest bias you can achieve: The power supply capability, the heatsink's dissipation ability and the SOA of the output devices.
With the F5T built according to the guide, the fourth limit is the point where the diode conducts, which is around 400mV. The diodes limit the voltage across the source resistor (or bypass the source resistor, if you choose to look at it another way) to 400mV, hence with them in circuit you will never be able to achieve a reading higher than that even though the amplifier can be running at a higher bias.
In general the safe envelope for each device is around 40W continuous dissipation, though they will be happier at lower operating points. It is also unwise to leave the diodes conducting at zero output because of bias runaway - so if you truly want a higher bias you may have to use lower values of source resistor, so that the conduction point occurs at higher currents. I prefer to add more pairs and run without diodes, but that's probably just me.
And you should watch the power supply voltage, you definitely don't want any sag. Heatsink temperatures look to be under control but 20 degrees difference between the devices and heatsink is an issue though, poor thermal contact maybe?
Thanks for your reply sangram.
So I understand that with the configuration of components that I have I can not go beyond 40 w output power for my speakers. How that translates feeding 4 ohm speakers?
If I put a resistor in parallel with 5 ohms with the two of 1 ohm, I low to resistance 0.45 ohm and bias rises to almost 0.9 Amp according to the formula and then if I get to 50w (3,55X3,55X8 /2=.50,5 w. is that correct?
You say you can not have more than 400mV readings, but I can raise the reading on the multimeter to 420mv, but we only have so left a couple of minutes for fear of seeing flames and smoke.
The difference in degrees between the heatsink and mosfet is 20 degrees. I used the gray insulation pad. I do not remember the material but it was not cheap.
So I understand that with the configuration of components that I have I can not go beyond 40 w output power for my speakers. How that translates feeding 4 ohm speakers?
If I put a resistor in parallel with 5 ohms with the two of 1 ohm, I low to resistance 0.45 ohm and bias rises to almost 0.9 Amp according to the formula and then if I get to 50w (3,55X3,55X8 /2=.50,5 w. is that correct?
You say you can not have more than 400mV readings, but I can raise the reading on the multimeter to 420mv, but we only have so left a couple of minutes for fear of seeing flames and smoke.
The difference in degrees between the heatsink and mosfet is 20 degrees. I used the gray insulation pad. I do not remember the material but it was not cheap.
No, that's not accurate. You can get more than 40W in Class A, but what you can read off the bias resistors is limited by the diodes. Normally Class A power halves and total power increases on a 4 ohm versus 8 ohm load, assuming a perfect amplifier. You can manipulate the 4 ohm Class A figures by overbiasing the amplifier, heatsinks permitting.
1.8A of total bias is sufficient for 50W in Class A, yes.
The 400mV is a base figure. It varies from device to device and with the temperature of the device. 5% is well within the region of variation normally associated with diode threshold voltages. How far you are willing to push depends on your risk profile.
If you can link to the material you used, it would help arrive at a reason why. 20 degrees for (say) 25 watts of power input means there is a possiblility of significant optimisation in the thermal interface.
Thanks again for your help!
I can not handle and remove the diodes, I have no knowledge to that. The configuration I have is that of the guide 6l6 and tables Diyaudio store .. I just want to find the maximum power with that configuration.
I think I will be conservative at the moment and leave a bias of 390 mV for a while, until the summer ... in Seville is hot !. Let's see how it goes with my 4 ohm speakers.
If lowering the resistance source get more power, maybe I'll try ...
I do not remember where I bought insulators transistor. Maybe it was at a local store and are silicone. What is the maximum temperature difference between the transistor and heat sink that can be accepted ?.
There are insulators keratherm In the diyaudio store, will I get more thermal bonding?
Bes regards.
I can not handle and remove the diodes, I have no knowledge to that. The configuration I have is that of the guide 6l6 and tables Diyaudio store .. I just want to find the maximum power with that configuration.
I think I will be conservative at the moment and leave a bias of 390 mV for a while, until the summer ... in Seville is hot !. Let's see how it goes with my 4 ohm speakers.
If lowering the resistance source get more power, maybe I'll try ...
I do not remember where I bought insulators transistor. Maybe it was at a local store and are silicone. What is the maximum temperature difference between the transistor and heat sink that can be accepted ?.
There are insulators keratherm In the diyaudio store, will I get more thermal bonding?
Bes regards.
For a 0.5C/W insulator, you would expect to see a 15Cdegree temperature difference, at 30W of dissipation.
I would reduce the 400 to 420mV of bias to around 300mV and remeasure your temperatures.
Then add another 1r0 in parallel to your existing 1r0||1r0 to give a new Rs of 0r333
With the same 300mV of bias you would now have 0.9A of bias per output pair.
2pair woould give you 1.8A of total bias for a maximum of 50W of ClassA into 8ohms, but only 25W of ClassA into 4ohms.
I would reduce the 400 to 420mV of bias to around 300mV and remeasure your temperatures.
Then add another 1r0 in parallel to your existing 1r0||1r0 to give a new Rs of 0r333
With the same 300mV of bias you would now have 0.9A of bias per output pair.
2pair woould give you 1.8A of total bias for a maximum of 50W of ClassA into 8ohms, but only 25W of ClassA into 4ohms.
Ok, their tips are a luxury, I'll get down. the work, but allow me a couple of questions:
if we reduce bias and resistance then won best temperature and better bias ... why does not arise so from the beginning? Is there any benefit in raising the bias, temperature and resistance originally?.
If 4 ohm speakers I have 25w class, the rest is up to 50w AB ?. I thought a folding amplifier power when the speaker impedance is reduced
Best Regards
if you reduce the bias voltage across the source resistor, you also reduce the bias current through the transistors.
This make it all run cooler.
Try that first.
Once you are happy that all is working OK after a few/many cold starts, then you can add the extra parallel 1r0 to each source resistor to reduce your rxisting 1r0||1r0=0r5 down to 0r3333
That increases the bias current and increases the heat. Try that. You may be happy with this increased heat, or you may want to reduce the bias voltage a little lower.
Then do some more cold starts and see how consistently it restarts and comes up to final quiescent currents and voltages.
This make it all run cooler.
Try that first.
Once you are happy that all is working OK after a few/many cold starts, then you can add the extra parallel 1r0 to each source resistor to reduce your rxisting 1r0||1r0=0r5 down to 0r3333
That increases the bias current and increases the heat. Try that. You may be happy with this increased heat, or you may want to reduce the bias voltage a little lower.
Then do some more cold starts and see how consistently it restarts and comes up to final quiescent currents and voltages.
P=IV=I²R=V²/R
when you reduce the load from 8ohms to 4ohms the power goes up.
It goes up because the current goes up.
But ClassA current has a fixed maximum value.
The ClassA power cannot go up when you reduce the load impedance
P=I²R for I=3.6Apk (=2*Ibias) and R= 8r0 the power is 3.6*3.6*8/2 (the 2 converts the pk values of sinewave to average levels for power).
If you reduce your load to 4 ohms the ClassA power drops to half.
when you reduce the load from 8ohms to 4ohms the power goes up.
It goes up because the current goes up.
But ClassA current has a fixed maximum value.
The ClassA power cannot go up when you reduce the load impedance
P=I²R for I=3.6Apk (=2*Ibias) and R= 8r0 the power is 3.6*3.6*8/2 (the 2 converts the pk values of sinewave to average levels for power).
If you reduce your load to 4 ohms the ClassA power drops to half.
I'm about 3/4th the way through building a F5 turbo V2 in a 5U chassis from the diyaudio with CLC filtering, 32V rails, cascoded the JFETs, Toshiba outputs. I've read a few hundred pages of this thread as well as others so I'm hoping it will go smoothly! Power supply is mostly done, FE boards and most of the output are soldered, just waiting on the Toshiba's to come in so I can finish off the outputs and do final wiring. It's coming out a LOT nicer than my first F5 which used a sock (clean of course) as an "insulator" for the rectifier diodes.
I have a question regarding the mounting the diodes on the output boards to the same heatsink as the output mosfets. According to the F5 Turbo article 300mV across the bias resistors is safe amount of bias where the diodes are "stable" (ie no thermal runaway), and 400mV is the point that potentially can cause them go into thermal runaway. I presume that observation is based on having them mounted on the same heatsink as the output which means even if they are they're at least at 55C.
Since the diodes conduct more based on their temperature, I was wondering if mounting them off the big heatsink with their own little heatsinks on each device would allow for higher biases before the diodes start really conducting and going into thermal runaway.
I'm really not sure how warm the diodes would get in this configuration since I don't really fully understand how the darn things work in this circuit. I'm assuming the great majority of the heat on the main sinks is coming from the transistors and heating the diodes, so isolating on their own sinks may keep them cooler. Since they're parallel to some 3W resistors then presumably they really aren't dissipating a lot of heat, correct?
Has anyone tried this and measured the temperature of the diodes to see how hot they run on their own sinks?
Brian
I have a question regarding the mounting the diodes on the output boards to the same heatsink as the output mosfets. According to the F5 Turbo article 300mV across the bias resistors is safe amount of bias where the diodes are "stable" (ie no thermal runaway), and 400mV is the point that potentially can cause them go into thermal runaway. I presume that observation is based on having them mounted on the same heatsink as the output which means even if they are they're at least at 55C.
Since the diodes conduct more based on their temperature, I was wondering if mounting them off the big heatsink with their own little heatsinks on each device would allow for higher biases before the diodes start really conducting and going into thermal runaway.
I'm really not sure how warm the diodes would get in this configuration since I don't really fully understand how the darn things work in this circuit. I'm assuming the great majority of the heat on the main sinks is coming from the transistors and heating the diodes, so isolating on their own sinks may keep them cooler. Since they're parallel to some 3W resistors then presumably they really aren't dissipating a lot of heat, correct?
Has anyone tried this and measured the temperature of the diodes to see how hot they run on their own sinks?
Brian
The temperature/Vf/If graph of the diodes is based on junction temperature, not surface temperature.
In a case when the diode is not conducting, the internal temperature may be lower than the case temperature. In fact, it is almost certain that this will be the case.
Hence the measurement of case temperature to estimate an optimal bias point may be hit and miss. When the case is closed, the temperature on any little heatsinks will float up to just a few degrees below the heatsink temperature, so even that measurement is not a definite insight into the junction temp of the diodes.
Hence the recommendation of the idling bias to be kept at 300mV-350mV or less, or to try it out with a very hot amplifier (as recommended in the documentation). That is really the only way to be sure that it will never run away in normal use.
In a case when the diode is not conducting, the internal temperature may be lower than the case temperature. In fact, it is almost certain that this will be the case.
Hence the measurement of case temperature to estimate an optimal bias point may be hit and miss. When the case is closed, the temperature on any little heatsinks will float up to just a few degrees below the heatsink temperature, so even that measurement is not a definite insight into the junction temp of the diodes.
Hence the recommendation of the idling bias to be kept at 300mV-350mV or less, or to try it out with a very hot amplifier (as recommended in the documentation). That is really the only way to be sure that it will never run away in normal use.
Alright, it's probably not worth the trouble to get a bunch of small heatsinks for each diode just to save a few deg C. I'm still curious if anyone has tried this since the Vf/If of the diodes are so highly influened by temperature. It seems the bias current would be less prone to runaway if the diodes didn't track to the temperature of the mosfets when they really start heating up.
My other question was about R25/R26. I cascoded the input jfets in case I ever want to go to a higher rail voltage. The schematic in the NP article seems to have an error that R25 is 10K and R26 (it's mirror on the N side) is 47.5K. I'd think they should be the same value. Is this correct?
Since I'm using a lower rail voltage (~32V) than the v3 which depicts the cascode @ 50V, should the R25/26 values be tweaked a little bit to give a higher voltage to the JFETs? What would be a good value?
Thanks for your help.
The ratio of the resistors should be approximately the same as the ratio of the supply voltage to the desired Vds across the JFETs. In the datasheet the JFETs enter low noise region below 12V Vds and this is a good target, but note that lower Vds also means higher input capacitance. The JFETs are safe to 25V or so, so there is plenty of room to fool around.
So if you have 48V on the supply the ratio of the resistors should be 3:1, if you have 36V it should be 2:1, and so on. In practice the voltage at the emitter will be 0.6V different from your chosen ratio. E12 series values work fine, minor inaccuracy in the ratio won't hurt. The mirrored values should obviously be the same, though I don't have the exact schematic that you are looking at (it's been changed around a bit). If you have spotted a likely error, it most probably is one. A picture would help.
As to the diodes, don't let me discourage you. The idea of this design concept is to experiment and learn. Feel free to do so
The warnings are only so you don't blow stuff up - the diodes make that a bit easier, hence the numerous posts on what they do and how to make the most of them while staying safe.
So if you have 48V on the supply the ratio of the resistors should be 3:1, if you have 36V it should be 2:1, and so on. In practice the voltage at the emitter will be 0.6V different from your chosen ratio. E12 series values work fine, minor inaccuracy in the ratio won't hurt. The mirrored values should obviously be the same, though I don't have the exact schematic that you are looking at (it's been changed around a bit). If you have spotted a likely error, it most probably is one. A picture would help.
As to the diodes, don't let me discourage you. The idea of this design concept is to experiment and learn. Feel free to do so
The warnings are only so you don't blow stuff up - the diodes make that a bit easier, hence the numerous posts on what they do and how to make the most of them while staying safe.mighty big car batteries.
I was was on a Narrow boat last week.
It had four 110Ahr 12V batteries. They were enormous.
Three were dedicated to the leisure side.
330Ahr @ 12V is roughly 49Ahr @ ±40Vdc, if the inverter was 100% efficient.
These three enormous batteries would last <10hours with a pair of ClassA amplifiers running at 2.5A bias.
I was was on a Narrow boat last week.
It had four 110Ahr 12V batteries. They were enormous.
Three were dedicated to the leisure side.
330Ahr @ 12V is roughly 49Ahr @ ±40Vdc, if the inverter was 100% efficient.
These three enormous batteries would last <10hours with a pair of ClassA amplifiers running at 2.5A bias.
Rixsta, you should just get an army of hamsters running on wheels to provide the AC. Less switching noise, although cleaning up the poop might get to be a chore.
I powered on one channel of my v2 build this morning and everything went great. I used 10K for R25/26 and the cascode was putting out 9.3V to the Jfets. Might play with that a little later but good enough for now. The first channel biased up nicely. Let it cook for an hour and it's stable at 350mV across source resistor on both N/P channels and 0mV offset.
I'm using Toshiba 2SK1530/2SJ201 and they came in TO264 packages, which have a lot more surface area than TO246. This seems to be a nice bonus for thermal transfer. I'm using Kerafol and the sinks are at 54C on their warmest point (inside an used screw hole by the FETs) and 53C on top. The outside of the packages are only 2C above the sink temp.
Gotta go to work now but I'll try to get the other channel mounted tonight and hopefully fire it up tomorrow.
Brian
I powered on one channel of my v2 build this morning and everything went great. I used 10K for R25/26 and the cascode was putting out 9.3V to the Jfets. Might play with that a little later but good enough for now. The first channel biased up nicely. Let it cook for an hour and it's stable at 350mV across source resistor on both N/P channels and 0mV offset.
I'm using Toshiba 2SK1530/2SJ201 and they came in TO264 packages, which have a lot more surface area than TO246. This seems to be a nice bonus for thermal transfer. I'm using Kerafol and the sinks are at 54C on their warmest point (inside an used screw hole by the FETs) and 53C on top. The outside of the packages are only 2C above the sink temp.
Gotta go to work now but I'll try to get the other channel mounted tonight and hopefully fire it up tomorrow.
Brian
Forget a DC powered SMPS, just stack 4 batteries in series (or 6 if you want 36v rails...) and then connect the amp PSU as bipolar. (Yes, it will actually work, remember that ground is only where we decide it to be...)
4 times the current, no SMPS or inverter to get in the way an make inefficiency, and you'll have such copious amounts of instant current capability that any momentary short circuit will instantly turn that component into a fuse. (Including the speaker)
4 times the current, no SMPS or inverter to get in the way an make inefficiency, and you'll have such copious amounts of instant current capability that any momentary short circuit will instantly turn that component into a fuse. (Including the speaker)
