There is only one thing that is bothering me on this Cyrus one.
Even though the outputs don't get very hot (I have a pretty heafty heatsink in each cabinet) I can't adjust the bias for less than 50-60
milivolts across each of the 0.22 ohm resistors. This represents 250 mA
The 200 ohm pot doesn't go any further, it's at minimum and from there
it only goes to a higher setting. Both amplifiers behave the same way.
Does anybody have an explanation for this?
In the meantime the amplifiers play great and there is no trace of distortion. I didn't observe the effect of increasing the bias on the
scope, I have yet to do that. However, being a class B amplifier the
value of the bias seems to be a little high.
Not having an indication of the right bias value I'm a little puzzled,
may someone could make a little light on this.
Even though the outputs don't get very hot (I have a pretty heafty heatsink in each cabinet) I can't adjust the bias for less than 50-60
milivolts across each of the 0.22 ohm resistors. This represents 250 mA
The 200 ohm pot doesn't go any further, it's at minimum and from there
it only goes to a higher setting. Both amplifiers behave the same way.
Does anybody have an explanation for this?
In the meantime the amplifiers play great and there is no trace of distortion. I didn't observe the effect of increasing the bias on the
scope, I have yet to do that. However, being a class B amplifier the
value of the bias seems to be a little high.
Not having an indication of the right bias value I'm a little puzzled,
may someone could make a little light on this.
take it easy.jmateus said:
This is an 8ohm amplifier.
the SOAR shows that the severe 8ohm load is just outside the DC locus. This amp may just be able to cope with a 4 to 8ohm load if not driven hard and the heatsink remains cold. The curves just below SOA limit are for Tc=35degC.
The model predicts +-37Vdc (1.6Vripple above that) on the supply rails when driving an 8ohm load to 60W.
Winzip can't open this file. Compression method 9.
When I compress using normal the file exceeds the 100k limit (actually 117k).
How should I compress the file.
jmateus said:
The Cyrus 1 has a 'no load' rail voltage of just over +/-30V, the Cyrus 2 is a little higher and the Cyrus 2+PSX runs at +/-40V. From measurements taken on one of my Cyrus 1s, the rail voltage drops by about 5V at full load. This drop should be less on the Cyrus 2+PSX due to the higher VA rated transformer, dual (parallel) rectifier diodes and larger reservoir capacitors.
So, with your existing power supplies and the more rugged output transistors, you effectively have a monoblocked Cyrus 2+PSX. The PSX has two 15000uF capacitors per rail feeding both channels so you even have the same total reservoir capacitance!
jmateus said:
First read the schematic!
There is a note on the schematic that I attached to Post #17 which states that if you run out of adjustment on the preset reduce the value of R83.
I have mentioned previously, either in this thread or another one regarding the Cyrus, that the measured quiescent current of one of my Cyrus 1s was about 75mA. I cannot confirm that this is the correct value as I have not seen any service information for this amp.
In fact so it seems...Good!
However, the fact that I have 250 mA as the bias setting that I can't change might have to do with a higher supply voltage, but that I can't
change either unless I use a smaller voltage transformer, say 25 V
unrectified.
I'll see what happens if I increase the bias on the scope, even for a
few moments.
Thanks for the info.
However, the fact that I have 250 mA as the bias setting that I can't change might have to do with a higher supply voltage, but that I can't
change either unless I use a smaller voltage transformer, say 25 V
unrectified.
I'll see what happens if I increase the bias on the scope, even for a
few moments.
Thanks for the info.
Hi,
I was put down for suggesting improvements to the One.
But, way back in posts19 & 20 this was discussed.
Don't increase the bias, it is already too high. Vre~=55mV.
Reduce Vre to what ever keeps the heatsinks cold and then listen for sound quality improvements as you increase Vre up to 15mV. But keep checking those sink temperatures.
I was put down for suggesting improvements to the One.
But, way back in posts19 & 20 this was discussed.
I suspect you ignored it, due to the comments made about my suggestions.
Don't increase the bias, it is already too high. Vre~=55mV.
Reduce Vre to what ever keeps the heatsinks cold and then listen for sound quality improvements as you increase Vre up to 15mV. But keep checking those sink temperatures.
jmateus said:
The current through the Vbe multiplier is set by a ccs and so is independant of the rail voltage. Your increased Iq will be due to the higher gain of your output transistors compared to the original high speed switching transistors (gain 8 to 15). This means that there will be smaller currents flowing through the driver transistors which will lower their Vbe resulting in the need for a lower bias voltage from the Vbe multiplier (which is why I added the note to the schematic, I didn't know at the time what output transistors you were using).
The Vbe of Q39 can also vary from one device to another which will affect the required resistor values in the Vbe multiplier circuit.
The total current in the driver transistor stages is not affected by output device gain - the output devices merely syphon off the current they require from that already flowing through the drivers. The issue is simply the variation in Vbe of the new output devices compared to the original design, and as Geoff pointed out that of the bias generator transistor also, as well as the drivers.
oops
It's r85 that needs to be reduced.
The bias voltage applied to the bases of the drivers is
[1+ {R85/R83+200r pot value}]*VbeQ39
using 650mV as the voltage for the multiplier transistor.
gives
max Vbias= 4.09V
min Vbias= 3.31V.
now looking at the 6 output transistors
Vre=[Vbias-6*0.65V]/2=4.09-3.9/2=95mV down to zero.
However when you add in a range of Vbe tolerances the minimum Vbias can be higher and minimum Vre can be as high as you have found.
Assuming worst case figures, R85 could be as low as 3k1, use 3k0.
It's r85 that needs to be reduced.
The bias voltage applied to the bases of the drivers is
[1+ {R85/R83+200r pot value}]*VbeQ39
using 650mV as the voltage for the multiplier transistor.
gives
max Vbias= 4.09V
min Vbias= 3.31V.
now looking at the 6 output transistors
Vre=[Vbias-6*0.65V]/2=4.09-3.9/2=95mV down to zero.
However when you add in a range of Vbe tolerances the minimum Vbias can be higher and minimum Vre can be as high as you have found.
Assuming worst case figures, R85 could be as low as 3k1, use 3k0.
Andrew, you beat me to it!. I have just returned to the forum to say that if the Iq needs to be reduced then, of course, R83 must have a higher value than 680R (or, as you correctly point out, R85 should be reduced in value).
Simulating the circuit with a variety of transistor models indicated that 680R plus a 200R preset should cover the majority of situations and that it was more likely that the Iq could not be set high enough so R83 would need to be lower (as you also pointed out earlier in the thread).
I did not envisage a situation arising where the minimum Iq that could be set was too high so my apologies, John, for any inconvenience caused.
Simulating the circuit with a variety of transistor models indicated that 680R plus a 200R preset should cover the majority of situations and that it was more likely that the Iq could not be set high enough so R83 would need to be lower (as you also pointed out earlier in the thread).
I did not envisage a situation arising where the minimum Iq that could be set was too high so my apologies, John, for any inconvenience caused.
Hi,
I still think it better to reduce R83 to improve the range of voltages available from the multiplier.
Geoff,
I think your take on reduced current demand from high gain outputs will reduce the Vbe of the drivers. Thus requiring a lower Vbias from the multiplier.
If Y grade 3519 were used the gain @ 5A is 90 to 180. The gain at around 70mA could still be over 100.
The reduced current flowing to the output will reduce the total current flowing through the driver and so the Vbe must be lower.
This in turn will have a tiny reduction on Ic through the pre-driver.
I still think it better to reduce R83 to improve the range of voltages available from the multiplier.
Geoff,
I think your take on reduced current demand from high gain outputs will reduce the Vbe of the drivers. Thus requiring a lower Vbias from the multiplier.
If Y grade 3519 were used the gain @ 5A is 90 to 180. The gain at around 70mA could still be over 100.
The reduced current flowing to the output will reduce the total current flowing through the driver and so the Vbe must be lower.
This in turn will have a tiny reduction on Ic through the pre-driver.
AndrewT said:
No disagreement from me, though I think you meant "reduce R85". Alternatively, R83 could be reduced to say 560R and VR1 increased to 500R which should give an even wider range of adjustment.
How about lowering R83 still further and running the amp in Class-A (heatsinks permitting)?
I am assuming that the total current in the driver stages is more than that required by the output transistors. In that scenario, say 30mA is flowing in the driver stage. The output device may require say 4mA, which is syphoned off leaving the remaining 26mA to return to the output line in the case of the top half, or the -ve rail in the case of the bottom half.
Now if higher gain output transistors are fitted, the driver stage still sources 30mA, but if the output device requires only 3mA that leaves 27mA returning.
Now if higher gain output transistors are fitted, the driver stage still sources 30mA, but if the output device requires only 3mA that leaves 27mA returning.
richie00boy said:
Just taking the upper half of the circuit as an example, the current through the driver transistor (Q31) is the sum of the current through R93 and the base current of output transistor (Q41).
The current through R93 is determined by the Vbe of Q41 and the base current of Q41, for a given collector current, depends on the gain of Q41.
So, there is no 'syphoning off' as you put it. The current through the driver transistor, and hence its Vbe, will change with variations in both the Vbe and the gain of the output transistor.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.