Dennis & Dirk -
As much as I would like to blame P1 & P2 for the failure (part failure), I think it was my fault. I'm building mono block Balanced versions of the F5t v3 with power supplies in separate enclosures.
My rail voltage was 51 vdc (no load), as soon as I had 200 mV of bias, the rail voltage was 48.1 vdc. When adjusting bias, P2 made the difference, and yes, you have to adjust VERY slow, no more than 3 - 4 degrees of a turn gives 10 - 20 mV of bias increase.
My problem was DC offset was never stable and P1 did not seem to be able to stabilize it regardless of turns of the pot.
I think the issue is, I'm measuring DC offset incorrectly. I have the P & N channels output together and the P & N channels on the other side tied together. I've been measuring DC offset between speaker outputs, I should be measuring DC offset between N and P channels.
My attempt to get any kind of result from P1 to stabilize DC offset caused P2 to fail. Instead of both P1 & P2 adjustments being a close number of turns increase, I was probably way off.
I have another mono block to adjust, so before I go above 200 mV bias, I'll measure DC offset between P and N channel outputs. First I need to swap out R5 and R6 on both FE boards.
As much as I would like to blame P1 & P2 for the failure (part failure), I think it was my fault. I'm building mono block Balanced versions of the F5t v3 with power supplies in separate enclosures.
My rail voltage was 51 vdc (no load), as soon as I had 200 mV of bias, the rail voltage was 48.1 vdc. When adjusting bias, P2 made the difference, and yes, you have to adjust VERY slow, no more than 3 - 4 degrees of a turn gives 10 - 20 mV of bias increase.
My problem was DC offset was never stable and P1 did not seem to be able to stabilize it regardless of turns of the pot.
I think the issue is, I'm measuring DC offset incorrectly. I have the P & N channels output together and the P & N channels on the other side tied together. I've been measuring DC offset between speaker outputs, I should be measuring DC offset between N and P channels.
My attempt to get any kind of result from P1 to stabilize DC offset caused P2 to fail. Instead of both P1 & P2 adjustments being a close number of turns increase, I was probably way off.
I have another mono block to adjust, so before I go above 200 mV bias, I'll measure DC offset between P and N channel outputs. First I need to swap out R5 and R6 on both FE boards.
elwood625 -
Thanks for your explanation. It may help others that want to try the balanced version.
Learning alongside. Is it correct that you were biasing one phase at a time (one set of boards on one heatsink) while measuring the DC offset between each output between the two sets of boards?
Phrased another way, were you not measuring offset relative to GND on the boards you were biasing, but between the two outputs of the two separate sets of boards as they'd be wired when connected to the speakers when bridged?
I think ZM referred to the offset relative to GND for one phase as "relative offset" and the offset between the boards between the speaker outputs across the two phases as "absolute offset". I may have that backwards... I always forget.
Either way, I think I understand what happened.
Again, so sorry that happened, but I really appreciate you posting it.
Thanks for your explanation. It may help others that want to try the balanced version.
Learning alongside. Is it correct that you were biasing one phase at a time (one set of boards on one heatsink) while measuring the DC offset between each output between the two sets of boards?
Phrased another way, were you not measuring offset relative to GND on the boards you were biasing, but between the two outputs of the two separate sets of boards as they'd be wired when connected to the speakers when bridged?
I think ZM referred to the offset relative to GND for one phase as "relative offset" and the offset between the boards between the speaker outputs across the two phases as "absolute offset". I may have that backwards... I always forget.
Either way, I think I understand what happened.
Again, so sorry that happened, but I really appreciate you posting it.
Hi guys, reporting back some information/checks.
Andy, I double checked the orientation of the capacitors and they are correct.
Dennis,
I checked voltages and this is what I came up with reading across the resistors:
+ side of fe
R29=-.8vdc R25=+27.5vdc R27=+4.4 R5=0vdc?
- side of fe
R28=+2.7 R6=0vdc R26=-29.5 R30=-1vdc
DC offset is .5vdc
My rail voltage at the board is +34.5 & -34.4vdc
Last edited:
Probably not related but looking at the photos in your initial posts, the V+ seems very close to the metal nuts holding the boards and I can see tiny wires of it spreading. Might cause a future issue in my opinion.
I ran the two F5T at 36.5V cascoded (under load) and they get to about 52.5C after 1 hour or so when ambient is 24C at 300mV bias. Absolute offset starts negative on both channels at around -350mVdc and when the amps get to around 50C settles in the 1-6mVdc range remaining stable (+/- 2mVdc depending on grid voltage fluctuations).
I ran the two F5T at 36.5V cascoded (under load) and they get to about 52.5C after 1 hour or so when ambient is 24C at 300mV bias. Absolute offset starts negative on both channels at around -350mVdc and when the amps get to around 50C settles in the 1-6mVdc range remaining stable (+/- 2mVdc depending on grid voltage fluctuations).
to all F5T-builders
Hello F5T builders,
if you increase your rail-voltages (especially if you go very high):
- you shouldn't go over the maximum dissipation those Mosfets can handle!
normally you have to decrease your bias
(example: my F5T runs on +-42,4 V rails, Case/heatsinks are 500mm deep / 210mm high for 2 n-channel boards = 4 Mosfets+4 Diodes; the same heatsinks
for 2 p-channel boards - Monoblocks!
I can't bias higher than 330mV per Mosfet and I have around 52° C at my
heatsinks).
So if you run railvoltages of +-51V you will reach the maximum heat those Mosfets can handle at a lower bias-point.
I made a screenshot of a part of Nelson Pass article about the F5T.
See pic.
I am assuming, that some of you want to 'barbecue' their Mosfets?
Greets
Dirk
Hello F5T builders,
if you increase your rail-voltages (especially if you go very high):
- you shouldn't go over the maximum dissipation those Mosfets can handle!
normally you have to decrease your bias
(example: my F5T runs on +-42,4 V rails, Case/heatsinks are 500mm deep / 210mm high for 2 n-channel boards = 4 Mosfets+4 Diodes; the same heatsinks
for 2 p-channel boards - Monoblocks!
I can't bias higher than 330mV per Mosfet and I have around 52° C at my
heatsinks).
So if you run railvoltages of +-51V you will reach the maximum heat those Mosfets can handle at a lower bias-point.
I made a screenshot of a part of Nelson Pass article about the F5T.
See pic.
I am assuming, that some of you want to 'barbecue' their Mosfets?
Greets
Dirk
Attachments
Hi guys, any thoughts on this below? I’ve included a pic of the temp of one of the fets after 30 seconds.
Quote:
Originally Posted by Dennis Hui View Post
luvrockin,
I may have missed it but do you have any voltage measurements you can post?
I'm not familiar with the F5t but looking at the schematics I would
suggest measuring the voltages across R25, R26, R27, R28, R3, R4, R5, R6 as well as the DC offset.
With P1/P2 low and the output stage off you may see the cascode transistors to run a little bit
hotter, both from the unloaded rails, but also from higher voltage across the collector
and emitter. But you should still be dealing with a fraction of a watt.
Hi guys, reporting back some information/checks.
Andy, I double checked the orientation of the capacitors and they are correct.
Dennis,
I checked voltages and this is what I came up with reading across the resistors:
+ side of fe
R29=-.8vdc R25=+27.5vdc R27=+4.4 R5=0vdc?
- side of fe
R28=+2.7 R6=0vdc R26=-29.5 R30=-1vdc
DC offset is .5vdc
My rail voltage at the board is +34.5 & -34.4vdc
Quote:
Originally Posted by Dennis Hui View Post
luvrockin,
I may have missed it but do you have any voltage measurements you can post?
I'm not familiar with the F5t but looking at the schematics I would
suggest measuring the voltages across R25, R26, R27, R28, R3, R4, R5, R6 as well as the DC offset.
With P1/P2 low and the output stage off you may see the cascode transistors to run a little bit
hotter, both from the unloaded rails, but also from higher voltage across the collector
and emitter. But you should still be dealing with a fraction of a watt.
Hi guys, reporting back some information/checks.
Andy, I double checked the orientation of the capacitors and they are correct.
Dennis,
I checked voltages and this is what I came up with reading across the resistors:
+ side of fe
R29=-.8vdc R25=+27.5vdc R27=+4.4 R5=0vdc?
- side of fe
R28=+2.7 R6=0vdc R26=-29.5 R30=-1vdc
DC offset is .5vdc
My rail voltage at the board is +34.5 & -34.4vdc
Attachments
I can’t comment on the voltage measurement. But of that temp is 194 degrees fahrenheit on your cascode transistors, or any transistor for that matter, SWITCH THE AMP OFF NOW, my friend. It would be nice to continue this discussion while you still have your eyes and hands intact.
A puzzle this, looking forward to comments on your measurements. You’ll solve this, man!
A puzzle this, looking forward to comments on your measurements. You’ll solve this, man!
Not nescessarily the reason. But i see some of your diodes are independently sinked. I cite papa:
« The point at which the diodes conduct is temperature dependent, so you will want to set the bias so that it makes a nice transition above the bias point and doesn't run away when the amplifier gets hot.»
So, I recommend you rework your sinks so all diodes and transistors have the same temp working conditions. Meaning all of them mounted to the main heatsink.
Also, I can’t say for sure, but it sorta looks like your fets were soldered to the boards prior to mounting to sink. Can you shead some light on whether that is so or not?
If so, you run a high risk of the fets not making good and or equal contact with the sink. They should be tightened before soldering to board, at 1-1,2nm. Maybe you did all of this, sorry if so, but a possible tip that might also help explain the situation.
I also count a number of cold solder joints on the output boards. Not nescessarily a contributor to your current issue, but for longevity you should consider reflowing them.
Regards,
Andy
Last edited:
Lol! That’s why I’m hesitant to go any further. That was after 30 seconds. That was the longest I ever ran it but I wanted to get the point across. I believe I soldered in the transistors and then the sinks, but they float. The contact seems to be good. I have new cascode transistors ordered and should be here any day. The heat sinks are cheapies that I bought off eBay. I thought perhaps they’re not disapating the heat properly so I bought some others off Digikey today.
I got nothing. The voltages across your cascode resistors, should be roughly
2/3 of rail across the 10k and 1/3 of the rail voltage across 4.75k. It doesn't look like you
have that. You shouldn't be seeing significant voltage across R29 (475R) resistor
and yet you have 0.8R. Your temperature is *really* high and you are wise to switch
it off. There should only be a fraction of a watt through the cascode
transistors.
You're also seeing non-trivial dc offset despite R5 and R6 voltages being zero,
which should mean the output transistors are all off...
Sorry but I can't think of anything beside checking for shorts and soldering issues.
2/3 of rail across the 10k and 1/3 of the rail voltage across 4.75k. It doesn't look like you
have that. You shouldn't be seeing significant voltage across R29 (475R) resistor
and yet you have 0.8R. Your temperature is *really* high and you are wise to switch
it off.
There should only be a fraction of a watt through the cascode transistors.
You're also seeing non-trivial dc offset despite R5 and R6 voltages being zero,
which should mean the output transistors are all off...
Sorry but I can't think of anything beside checking for shorts and soldering issues.
Question on what could possibly be an issue with my F5 turbo. I added my turntable back into my system today, but before doing that, I changed the gain setting on my Pearl 2 in order to get some more gain out of my LOMC Denon DL-103. Everything was fine after an hour, so I decided to swap in my Dynavector 10x5 HOMC in not thinking it could be too much gain.
System played for about 10 minutes then the amp shut off. I realized I blew one of my 2.5A fuses but barely as I had to look closely to see the tiny separation. I then realized my gain could have been so high it caused excessive clipping and the fuse to blow. I did not have a spare 2.5A on me, so I added a 5A and plugged the F5 turbo into my DBT. It's been such a long time since I first built the amp I couldn't remember if the bulb was supposed to dim but it appeared to stay the same brightness on a 60 watt bulb but did not seem fully bright like the bulb was 3/4 lit. Capacitors may have still been charged so wondering if that is why it did not dim before brightening. I measured 6volts +- on the rails with the DBT...
I made sure my speakers did not get damage from the excessive gain/clipping and they played fine on another amp. I will pick up a couple more 2.5A fuses tomorrow but wanted to see if there could be another issue that may have arisen from excessive gain/clipping that I should be ready to trouble-shoot.
System played for about 10 minutes then the amp shut off. I realized I blew one of my 2.5A fuses but barely as I had to look closely to see the tiny separation. I then realized my gain could have been so high it caused excessive clipping and the fuse to blow. I did not have a spare 2.5A on me, so I added a 5A and plugged the F5 turbo into my DBT. It's been such a long time since I first built the amp I couldn't remember if the bulb was supposed to dim but it appeared to stay the same brightness on a 60 watt bulb but did not seem fully bright like the bulb was 3/4 lit. Capacitors may have still been charged so wondering if that is why it did not dim before brightening. I measured 6volts +- on the rails with the DBT...
I made sure my speakers did not get damage from the excessive gain/clipping and they played fine on another amp. I will pick up a couple more 2.5A fuses tomorrow but wanted to see if there could be another issue that may have arisen from excessive gain/clipping that I should be ready to trouble-shoot.
The heatsink post was directed at Elwoods, not your build. Though, it never hurts taking a look at everything when there is trouble. But crappy heatsinks does not explain 194F. =S
Just a thought: is it possible you put the N-channel cascode transistor on the P channel side and vice versa? Maybe I asked this before, but possibly worth checking out.
Cheers,
Andy
Last edited:
I don't think we're at the boat anchor stage so don't despair yet.
You didn't measure the voltages across the 10 ohm source resistors R3 and R4.
I suspect they will be completely out of whack.
Can you check if the jfets (2sk170/2sj74) are in the correct spots? You will probably need to
cut off the zip tie you have on them and gently bend the jfets back.
You didn't measure the voltages across the 10 ohm source resistors R3 and R4.
I suspect they will be completely out of whack.
Can you check if the jfets (2sk170/2sj74) are in the correct spots? You will probably need to
cut off the zip tie you have on them and gently bend the jfets back.