....wow thanks for the reply the evul F5 is interest of mine since when you buy the diystore pcbs for a 2 mono bloc F5V3 you end up with 4 input boards and I happen to have 2 sets extra jfets maybe later on since I would need as you metion input transformer etc, for balanced signal input.....
...the extra output boards on a F5 V3 I actually have extra U5 heat sinks but need to find a case or mount solution before I could do that...
.....or replace the U5 chassis front aluminum plate with 2 heat sinks for the extra boards, not sure if the bracketing fits the front though and I kind of like the front aluminum panel.....
...the extra output boards on a F5 V3 I actually have extra U5 heat sinks but need to find a case or mount solution before I could do that...
.....or replace the U5 chassis front aluminum plate with 2 heat sinks for the extra boards, not sure if the bracketing fits the front though and I kind of like the front aluminum panel.....
Last edited:
That's the first time I've seen the math that way. It's usually written 32V Peak * 0.707, or 48Vpeak *0.707 because RMS is based on peak amplitude of a sine wave:
Difference Between Peak to Peak and RMS | Peak to Peak vs RMS
Yes power is written P=IV, but V can be written as V=IR, so P=I^2*R works just as well.
Consequently, in a pure Class A amp, so long as the amp can deliver twice the bias current, high voltage is not necessary.
EG.
My transformer is rated at total 2.4KVA, with 60V No Load secondaries. There are 5 pairs of secondaries, so 480 VA each. There are two primaries, each of which can be wired for European, Japanese or North American primary voltages. I'm guessing this transformer was intended for a Home Theatre amp being sold to the global market.
By wiring the higher voltage primaries in series, it would take European power. But if providing NA power to this wiring, a nominal 30 VDC is produced, which cuts the VA in half -- 240 VA each secondary pair.
One pair of secondaries is not used, the other 4 are split into two pairs, one per channel. So, 480 VA is available per channel.
This transformer, could be wired so as to increase the secondary voltage, but still be in a useful range. ( one primary Euro, the other NA or Japanese for example )
Is there any reason to increase the secondary voltage?
Not from a heatsinking perspective: If the bias current remains the same, the voltage drop across the output MOSfet will go up. So the power dissipated would increase. Meaning bigger heatsinks would be needed for the same temp rise at the same bias current.
Therefore, in this amp, lower rail voltages are better, as they lower the heat dissipation needs of the heat sinks.
Last edited:
6L6 used .3535 because he's using peak-peak volts (volts x2) instead of just volts. (.3535 x2 =.707)
The F5 is push pull so when it runs out of class A bias current (times ~2) it can keep going in class AB up to the voltage limit and/or current limit of the power supply. Unlike a single ended amp which is limited by bias current.
Simply put, in the F5, more voltage = more watts. As long as total current demand can be met by the power supply. There is a limit of course, the amount of total power the devices can reliably endure and how much your heat sinks can dissipate.
The limits are all explained in the F5 Turbo manual.
I use V^2 / Rload / 2 = ~RMS watts. It's just easier to calculate and is close enough....
Example, 46 volts peak and an 8 ohm load:
46 x 0.707 = 32.522 volts RMS -or- (46+46 = peak to peak) x 0.3535 = 32.522 RMS volts
32.522^2 / 8 = 132.2100605 RMS watts
or
46^2/8/2 = 132.25 RMS watts
TJ
The F5 is push pull so when it runs out of class A bias current (times ~2) it can keep going in class AB up to the voltage limit and/or current limit of the power supply. Unlike a single ended amp which is limited by bias current.
Simply put, in the F5, more voltage = more watts. As long as total current demand can be met by the power supply. There is a limit of course, the amount of total power the devices can reliably endure and how much your heat sinks can dissipate.
The limits are all explained in the F5 Turbo manual.
I use V^2 / Rload / 2 = ~RMS watts. It's just easier to calculate and is close enough....
Example, 46 volts peak and an 8 ohm load:
46 x 0.707 = 32.522 volts RMS -or- (46+46 = peak to peak) x 0.3535 = 32.522 RMS volts
32.522^2 / 8 = 132.2100605 RMS watts
or
46^2/8/2 = 132.25 RMS watts
TJ
To calculate how many watts in class A...
Using the previous example, 46 volts, 2.8a bias (.7a per pair x 4 pair), 8 ohm load:
Total RMS watts: 46^2/8/2 = 132.25
Peak class A volts = (Ibias x 2) x rload
so: (2.8a x 2) x 8 ohms = 44.8v peak
then: 44.8^2/8/2 = 125.44 RMS watts in class A
The remaining RMS watts would be in class AB
My maths are a little unorthodox but it works and is easy to remember.
TJ
Using the previous example, 46 volts, 2.8a bias (.7a per pair x 4 pair), 8 ohm load:
Total RMS watts: 46^2/8/2 = 132.25
Peak class A volts = (Ibias x 2) x rload
so: (2.8a x 2) x 8 ohms = 44.8v peak
then: 44.8^2/8/2 = 125.44 RMS watts in class A
The remaining RMS watts would be in class AB
My maths are a little unorthodox but it works and is easy to remember.
TJ
Takitaj,
The maths are clear: RMS is "ROOT MEAN SQUARED". It is the square root of the average value squared and has a clear derivation that is independent of peak to peak voltage.
Check the link I posted.
Squaring the average removes peak to peak voltage from consideration.
I don't recommend anyone thinking about RMS voltage in terms of peak to peak Voltage and 1/2 *1/sqrt(2).
While the numbers derived are right, the formula does not have the same physical significance.
The maths are clear: RMS is "ROOT MEAN SQUARED". It is the square root of the average value squared and has a clear derivation that is independent of peak to peak voltage.
Check the link I posted.
Squaring the average removes peak to peak voltage from consideration.
I don't recommend anyone thinking about RMS voltage in terms of peak to peak Voltage and 1/2 *1/sqrt(2).
While the numbers derived are right, the formula does not have the same physical significance.
The usual maths are:
1) IclassA = 2*Ibias
2) VclassA = IclassA*Rload (V=I*R)
3) PclassA = IclassA^2*Rload (P=I*V=I^2*R)
In the example above, that would be:
IclassA = 2*2.8 = 5.6A
VclassA = 5.6*8 = 44.8
PclassA = 5.6*44.8 = 5.6^2*8 = 250.88 peak
Note that PclassA is peak. RMS/Average is half: 125.44.
Higher voltages, up to rail voltage, are class AB. P=IV = V^2/R = 264.5 peak, 1/2 is RMS = 132.25
NOTE that your Heat sinking requirements would be taken using the RMS power.
1) IclassA = 2*Ibias
2) VclassA = IclassA*Rload (V=I*R)
3) PclassA = IclassA^2*Rload (P=I*V=I^2*R)
In the example above, that would be:
IclassA = 2*2.8 = 5.6A
VclassA = 5.6*8 = 44.8
PclassA = 5.6*44.8 = 5.6^2*8 = 250.88 peak
Note that PclassA is peak. RMS/Average is half: 125.44.
Higher voltages, up to rail voltage, are class AB. P=IV = V^2/R = 264.5 peak, 1/2 is RMS = 132.25
NOTE that your Heat sinking requirements would be taken using the RMS power.
Last edited:
Check your offset voltage. You will need to re adjust balance/offset again with a new set of mosfets. You may need to disable the speaker protection to set bias/offset. Turn down the pots for minimum bias current and redo the bias. After you have bias and offset where you want it you can re enable the speaker protection.
P1 and P2 were set to zero.
Protection went off on both channels.
I am using teabags power supply.
The led on the negative rail on the channel that got new Mosfets remained on after power was shut down. There was 15 volts in it, dropping slowly.
Reverified on both channels that the mosfets were isolated. To do this, I removed all connections to the front end and power supply. Ohms from output to case show Open Loop.
I then measured ohms from the back of each MOSFET to case. Also OL
Protection went off on both channels.
I am using teabags power supply.
The led on the negative rail on the channel that got new Mosfets remained on after power was shut down. There was 15 volts in it, dropping slowly.
Reverified on both channels that the mosfets were isolated. To do this, I removed all connections to the front end and power supply. Ohms from output to case show Open Loop.
I then measured ohms from the back of each MOSFET to case. Also OL
Last edited:
Well that's odd.... the speaker protector flashes on power up like it should. The relay makes and lights go solid, as they should.
Then with rails at +/-39 volts, there is +35 VDC on the output and the speaker protection did not trip.
When I measure the voltages across the source resistors, P1 and P2 are zero. However, there is .310 across a Pchannel source resistor test point, and 40VDC on the output.
Trying to decrease the output offset with the negative channel starts to work, but the voltage across the P-channel test point increases -- I saw 0.6VDC, when the output offset got down to about 24V. At the time the voltage across the N-channel test point was showing 0.165 or a bit more.
Could the trimmer that controls the P channel bias be broken have a poor connection?
Then with rails at +/-39 volts, there is +35 VDC on the output and the speaker protection did not trip.
When I measure the voltages across the source resistors, P1 and P2 are zero. However, there is .310 across a Pchannel source resistor test point, and 40VDC on the output.
Trying to decrease the output offset with the negative channel starts to work, but the voltage across the P-channel test point increases -- I saw 0.6VDC, when the output offset got down to about 24V. At the time the voltage across the N-channel test point was showing 0.165 or a bit more.
Could the trimmer that controls the P channel bias be broken have a poor connection?
Something is wrong on the + side of the circuit. With P1/P2 at 0 ohms there should not be any drive voltage available to turn on the mosfets.
The negative PS LED probably stayed on because there was no load on it to dischage the capacitors.
Also if you have the turbo diodes installed you should not be able to get .6 volts across them without a lot of current flowing. Not sure what is going on but you need to check all transistors/fets looking for shorts, check that ps and new fets were installed correctly etc.
The negative PS LED probably stayed on because there was no load on it to dischage the capacitors.
Also if you have the turbo diodes installed you should not be able to get .6 volts across them without a lot of current flowing. Not sure what is going on but you need to check all transistors/fets looking for shorts, check that ps and new fets were installed correctly etc.
No shorts to case for certain.
If there is that much current flowing, the FE board must be driving it. I will measure Vgate each side next.
At this point, I am cleaning up the FE board a little.
The resistances on the test points check fine, so the trimmers are installed OK. I measure a maximum of 840 Ohms on one and 810 on the other.
I don't know why they should be so different.
They both turn to zero, which implies no current should flow in the outputs.....
.
If there is that much current flowing, the FE board must be driving it. I will measure Vgate each side next.
At this point, I am cleaning up the FE board a little.
The resistances on the test points check fine, so the trimmers are installed OK. I measure a maximum of 840 Ohms on one and 810 on the other.
I don't know why they should be so different.
They both turn to zero, which implies no current should flow in the outputs.....
.
The issues:
1) Near full Positive rail DC offset with trimmers at zero
2) Increasing N-channel trimmer resistance increases P-Channel current.
2 is true at least on the output board I measured.... in my version, output boards are in parallel.
Minor changes from schematic on FE board. R25,R26 are 12K (not 10K). R27 R28 are 5.6K ( not 4.75K ) Ratio is .466, not .475. The base resistors to the cascode are 470 Ohms, not 475..
More measurements tomorrow.
Would there be any sense to apply power to the output boards by themselves? This would show beyond any doubt that the FE board is doing all the work, as when Vgate = zero on an output MOSfet, no current flows.
1) Near full Positive rail DC offset with trimmers at zero
2) Increasing N-channel trimmer resistance increases P-Channel current.
2 is true at least on the output board I measured.... in my version, output boards are in parallel.
Minor changes from schematic on FE board. R25,R26 are 12K (not 10K). R27 R28 are 5.6K ( not 4.75K ) Ratio is .466, not .475. The base resistors to the cascode are 470 Ohms, not 475..
More measurements tomorrow.
Would there be any sense to apply power to the output boards by themselves? This would show beyond any doubt that the FE board is doing all the work, as when Vgate = zero on an output MOSfet, no current flows.
Thanks!
I will check. Yes. All at 47.5 or thereabouts. One N channel wants to read about 48 sometimes.... hard to get the probes in.
Last edited: