If one "starts up" with a very low bias current to enable the SMPS to start up normally, then I would be very tempted to use a timer to activate the next stage of output bias.
The highest stage of output bias could be activated with a manual Summer/Winter switch, or could be done with a temp sensor and switched to Summer setting if the sink gets too hot. The comparator for the automated S/W switch over should have a high hysteresis Feed Back to prevent frequent back and forth switching.
The highest stage of output bias could be activated with a manual Summer/Winter switch, or could be done with a temp sensor and switched to Summer setting if the sink gets too hot. The comparator for the automated S/W switch over should have a high hysteresis Feed Back to prevent frequent back and forth switching.
Andrew,
I have a schematic that does all of that and quite a bit more. Including latching circuits for momentary switches, 555 timers for delays, hysterises on the comparators etc was all calculated. The layout was a bit too unwieldy by the time I added the rest of the functions that I wanted I could do it all for similar cost and have a more compact with a dedicated microprocessor. So I shelved that half done layout and will proceed on a hybrid where I replace parts of that into a microprocessor. I will ensure that the microprocessor and any remote comms are in deep sleep state unless necessary, they will make no difference to EMI/RFI near the amps.
I'll share what I come up with
Chris
I have a schematic that does all of that and quite a bit more. Including latching circuits for momentary switches, 555 timers for delays, hysterises on the comparators etc was all calculated. The layout was a bit too unwieldy by the time I added the rest of the functions that I wanted I could do it all for similar cost and have a more compact with a dedicated microprocessor. So I shelved that half done layout and will proceed on a hybrid where I replace parts of that into a microprocessor. I will ensure that the microprocessor and any remote comms are in deep sleep state unless necessary, they will make no difference to EMI/RFI near the amps.
I'll share what I come up with
Chris
Measuring & Setting Bias
Hi, another daft but important question... BIAS
Measuring: Multimeter in series with the Output V+ power supply & set to Ampere. As the trim pot is adjusted the current draw increases / decreases.
Setting: Bias required 360mA - MM in series, power on, trim pot set to minimum. Adjust trim pot until a stable 360mA is shown on MM...
Question: I take it that when it's said a bias of 360mA, that this is per rail? So we're getting a +/- 360mA (720mA) current draw from the supply.
Or... 360mA total, and we set 180mA per supply rail?
Hi, another daft but important question... BIAS
Measuring: Multimeter in series with the Output V+ power supply & set to Ampere. As the trim pot is adjusted the current draw increases / decreases.
Setting: Bias required 360mA - MM in series, power on, trim pot set to minimum. Adjust trim pot until a stable 360mA is shown on MM...
Question: I take it that when it's said a bias of 360mA, that this is per rail? So we're getting a +/- 360mA (720mA) current draw from the supply.
Or... 360mA total, and we set 180mA per supply rail?
All the above is correct and exactly describes the bias procedure. Remember that if you mounted the pots the way the silk layer indicates to, then minimum bias correlates to the pot being turned fully clockwise (all the way "up")
This is a commonly misunderstood area. Current doesn't add in series, it's the same throughout the loop. It flows (in the conventional sense) from the V+ terminal, through the N channel fet, then through the P channel fet, and back into the V- terminal completing the loop. The value you read on the multimeter is the current through this loop, and if you set it to say 360mA, then you'll measure the same value if you connect the DVMM between the V- supply and the P fet. The total current drawn from the supply is 360mA.
If you want to know the power dissipated by the output stage, then you need to consider the current through the loop (360mA) and the total voltage differential which is your |V+| + |V-|. If you have +/- 20V rails, then your total voltage would be 40V and your total dissipation in the output stage would be P=V*I or 40V * 360ma = 14.4W.
Regards,
Owen
Understanding it...
Brilliant, understood.
So using the DPS-600 +/-45/55 I get 90V * 0.360A = 32.4W
A higher Bias would equate to higher power dissipation & a need larger heatsinks.
In terms of Power Supply requirements, it should be capable of sustaining the current drawn by biasing, plus the required draw for music & head room?
Using a regulated power supply capable of 6 Amp [+/- 45V], as the bias increases, this decreases the amount left for powering the load & head room?
If bias is set to 1A and the remaining potential current draw is split 50/50, giving 2.5A to drive the load [8Ohm]. I should get 50W output?
However, at full output, the +/-45V power rails would cause a larger drain than the allocated 2.5A, the amp/supply would distort, clip, overheat, blow-up?
Paul
Brilliant, understood.
So using the DPS-600 +/-45/55 I get 90V * 0.360A = 32.4W
A higher Bias would equate to higher power dissipation & a need larger heatsinks.
In terms of Power Supply requirements, it should be capable of sustaining the current drawn by biasing, plus the required draw for music & head room?
Using a regulated power supply capable of 6 Amp [+/- 45V], as the bias increases, this decreases the amount left for powering the load & head room?
If bias is set to 1A and the remaining potential current draw is split 50/50, giving 2.5A to drive the load [8Ohm]. I should get 50W output?
However, at full output, the +/-45V power rails would cause a larger drain than the allocated 2.5A, the amp/supply would distort, clip, overheat, blow-up?
Paul
Exactly correct.
Not really... the bias current always contributes to the output current up until the point where the current delivered to the load exceeds the bias current, then the surplus required has to be delivered from the supply.
This also defines the point the amplifier leaves class A, as one of the output devices will stop conducting when twice the bias current is exceeded (for push-pull class AB). In this case, the amplifier will run in class A until the load requires more than 720mA peak. For an 8 ohm load, that means roughly 2 watts RMS.
It also means that the load on the power supply up until the 2W point is constant. The current either travels thought the transistors as bias current, or into the load as output power.
Again, not really. If you bias at 1A, the output stage will now operate in class A up to 2A peak of output current. For an 8 ohm load, that's roughly 16W without leaving class A.
If you want to know what your power supply limited output power would be, then you need to look at your peak output voltage, and your load, and figure out what your peak output current will be.
If you have 45V rails on the fets, and 55V rails on the front end, you should be able to swing within a few volts peak of the rails. Let's be conservative and say you can output 41V peak. With an 8 ohm load, that gives a peak output current of 5.125A. Your supply has to be able to provide that level of current, otherwise your rails will sag and your output will voltage clip earlier than the estimated 41V.
The above scenario would have a power output of 112 watts RMS into an 8 ohm load.
Note that this is regardless of the bias current. If you set the bias current to 2.5A, then your output power capability remains the same, as long as your supply can provide the needed peak output current.
In the case of the DPS600, you cannot bias at super high levels, as the continuous power draw from the supply will cause it to heat up significantly and shut down (or explode) due to thermal overload. Even though the supply is rated at 6A output current, it cannot do this continuously without getting too hot.
Regards,
Owen
I thought the SMPS was rated at a continuous 600W.
From +-45V that would equate to just over 6A continuous.
If we were to take the continuous 6A as the current limit of the DPS600 then it should be able to hold the regulated +-50Vdc at upto 6A continuous.
Or is the DPS600 rated at lower than 600W continuous?
peak current demand of a speaker is worse than the resistively modeled current shown above.
A single speaker driver is likely to demand at least 50% more than this on fast transients.
A passive crossover speaker is likely to demand at least 100% more and can be upto 5times the resistive current demand for fast transients.
I design for a peak current demand of 3times the resistive model, i.e. 8ohms requires 15Apk.
Andrew,
I expect the thermal limit of those SMPS is dependent on the mounting arrangement for thermal reasons.
Roberto posted this on the DPS600 thread:
From the datasheet that I have there are also the following hard protection limits in place:
- Output to a short circuit
- 14Amps drawn for longer than 100ms
The heat spreader provided on the DPS600 has mounting holes. Can be mounted to a heat sink, floor of enclosure or separate aluminium sheet. Roberto recommended 70mm wide x 5mm thick alu sheet for mounting 2 x DPS600. For mine I am using two separate 50mm wide x 3mm thick alu sheets spaced off the floor with plenty of ventilation holes above and below.
I think you'd come against a point where the thermal interface between the heat bar on the DPS600 and the alu sheets is limiting the thermal dissipation of the whole unit, this would become the limit for the continuous current rating. Though we don't have enough data available to do this other than by testing at different bias settings and measuring temperatures observed. This is no different to the limitation for a linear regulated power supply continuous rating, is it?
I expect that Owen was just using some intuition with the continuous current limitation. With 6A bias wouldn't the output stage heatsinking make the DPS600 a false economy in saved space? Remember that these PSU's have a fixed output voltage so you're dissipating for 45 or 55V (depending on the version purchased). So somewhere upwards of 400W idle thermal load??
I expect the thermal limit of those SMPS is dependent on the mounting arrangement for thermal reasons.
Roberto posted this on the DPS600 thread:
Hi, in this all..dps-600 break at >82° in side of Power-Unit. just a info
From the datasheet that I have there are also the following hard protection limits in place:
- Output to a short circuit
- 14Amps drawn for longer than 100ms
The heat spreader provided on the DPS600 has mounting holes. Can be mounted to a heat sink, floor of enclosure or separate aluminium sheet. Roberto recommended 70mm wide x 5mm thick alu sheet for mounting 2 x DPS600. For mine I am using two separate 50mm wide x 3mm thick alu sheets spaced off the floor with plenty of ventilation holes above and below.
I think you'd come against a point where the thermal interface between the heat bar on the DPS600 and the alu sheets is limiting the thermal dissipation of the whole unit, this would become the limit for the continuous current rating. Though we don't have enough data available to do this other than by testing at different bias settings and measuring temperatures observed. This is no different to the limitation for a linear regulated power supply continuous rating, is it?
I expect that Owen was just using some intuition with the continuous current limitation. With 6A bias wouldn't the output stage heatsinking make the DPS600 a false economy in saved space? Remember that these PSU's have a fixed output voltage so you're dissipating for 45 or 55V (depending on the version purchased). So somewhere upwards of 400W idle thermal load??
Wow... that's a lot of amps & a lot of heat to disperse under high biasing. Reading a few articles about regulated supply design & demands from various sources online, it came across that most aim at a providing constant current draw of 1/3 peak (or RMS) and let the Caps deal with the peak load.
However, the better way looks to have the supply being able to provide 100% peak or more, and having the Caps largely for smoothing!?
I could do with getting some proper books on the subject.
DPS-600, I understood that it could provide 6A constant draw, but had 13A short term in my head. Mine are mounted on a single aluminium sheet 250mm x 250mm x 3mm and will be raised 10mm off the chassis base.
I've been looking at supply needs as a precaution against my mishandling of the DPS-600 units I have, plus for other projects.
However, the better way looks to have the supply being able to provide 100% peak or more, and having the Caps largely for smoothing!?
I could do with getting some proper books on the subject.
DPS-600, I understood that it could provide 6A constant draw, but had 13A short term in my head. Mine are mounted on a single aluminium sheet 250mm x 250mm x 3mm and will be raised 10mm off the chassis base.
I've been looking at supply needs as a precaution against my mishandling of the DPS-600 units I have, plus for other projects.
Amp's Mute Function
More Questions:
I know there was a discussion about audible on/off thumps etc. during power-up/down... In case of a minor incident, will having the LME muted during these stages stop anything being passed through to the speaker?
Currently the Mute Function goes into Play Mode as the LME's Vcc (55V on mine) reaches 12V(?). Reading the datasheet, I've seen the max Imute, and that it says Vmute can be powered from Vcc but no mention of a max V... Within the amp's mute circuit would it be possible to change Z1 from 12V to say 39V or 43V, so that switching to/from Play Mode occurs once the rails are much higher?
Obviously, the values of R38 & R53 would need to change in conjunction with Z1's Vz to keep near the current Imute value... Would C75 also need changing to accommodate the higher V's?
This on top of a simple switch, when I want to mute play back.
More Questions:
I know there was a discussion about audible on/off thumps etc. during power-up/down... In case of a minor incident, will having the LME muted during these stages stop anything being passed through to the speaker?
Currently the Mute Function goes into Play Mode as the LME's Vcc (55V on mine) reaches 12V(?). Reading the datasheet, I've seen the max Imute, and that it says Vmute can be powered from Vcc but no mention of a max V... Within the amp's mute circuit would it be possible to change Z1 from 12V to say 39V or 43V, so that switching to/from Play Mode occurs once the rails are much higher?
Obviously, the values of R38 & R53 would need to change in conjunction with Z1's Vz to keep near the current Imute value... Would C75 also need changing to accommodate the higher V's?
This on top of a simple switch, when I want to mute play back.
Last edited:
Hi Paul,
I don't think the mute in the LME chip will fix all of these cases that produce spurious noises during power off. I found some of the posts where this was discussed previously and here is a few of the ones that I think clarified the issues:
And this might give you some more food for thought on options for the mute circuit.
@ hochopeper: Yep, those are some of the post that I was thinking of and had read. Shows that full Vcc can be used for Vmute, though from TI's reference docs a lower Vmute looks to be a preference.
One TI doc (snaa057a) for a proposed +/- 72V unregulated power supply for the LME49810/11/30, uses a similar mute circuit to opc's design, but sets Vmute at 3.9V and 'Play Mode' switching at 54.9V using two zeners.
Interesting paragraph from the doc:
Just trying to think of simple options as a just in case, doesn't take away from the need to correctly shut down the amp & supplies.
Paul
One TI doc (snaa057a) for a proposed +/- 72V unregulated power supply for the LME49810/11/30, uses a similar mute circuit to opc's design, but sets Vmute at 3.9V and 'Play Mode' switching at 54.9V using two zeners.
Interesting paragraph from the doc:
Just trying to think of simple options as a just in case, doesn't take away from the need to correctly shut down the amp & supplies.
Paul
Hi All,
I read some comments just now (I have not seen activity or listening tests with the DPS-600 in this thread for a long time).
Some Info: All DPS-600 in this GB (i know used in class AB), and then were tested with a load of 2x360mA (360mA per rail), during the on / off and temperature produced. (this means, 720mA at 55V, seen from the side of the power-unit).
All had no spike while turn on / off.
With this constant load, mounting vertically to the DPS-600 on an aluminum base 180x40x3mm with an ambient temperature of 28 °, the temperature is balanced at 44 °, 16 ° delta heat is not bad for this type of smps I remember does not use a pwm.
-----------------------------------------------------
Many DPS-600 at + /-68V (not in this GB), allow you to get on with mosfet amps, 400w @ 4R without problems.
I do not agree to try and measure the smps, with constant resistive loads to see at full power. I accept this for only a few investigations
We use to active load (instrument similar to fast mosfet amplifier, driven by pattern from the PC, so you can see the exact behavior and many parameters, ideal when it works as an amplifier.
Regards
I read some comments just now (I have not seen activity or listening tests with the DPS-600 in this thread for a long time).
Some Info: All DPS-600 in this GB (i know used in class AB), and then were tested with a load of 2x360mA (360mA per rail), during the on / off and temperature produced. (this means, 720mA at 55V, seen from the side of the power-unit).
All had no spike while turn on / off.
With this constant load, mounting vertically to the DPS-600 on an aluminum base 180x40x3mm with an ambient temperature of 28 °, the temperature is balanced at 44 °, 16 ° delta heat is not bad for this type of smps I remember does not use a pwm.
-----------------------------------------------------
Many DPS-600 at + /-68V (not in this GB), allow you to get on with mosfet amps, 400w @ 4R without problems.
I do not agree to try and measure the smps, with constant resistive loads to see at full power. I accept this for only a few investigations
We use to active load (instrument similar to fast mosfet amplifier, driven by pattern from the PC, so you can see the exact behavior and many parameters, ideal when it works as an amplifier.
Regards
No, but surface is well flat. fixed well with 2x4MA.
this alluminum rectangle, we use with both dps-600/500 in the laboratory.
For mute on wire amp, a good solution can be a window comparator on vcc rail. if<(vcc-30%) then switch. is better add a costant=1Sec. just in case of unregulated power supply.
Last edited:
I have just installed my boards into the old carcass of my monoblock amps. I just "tacked" in the power connections etc for the moment because I will build a new chassis for the amps and use them with DPS 500 power supplies I have ordered from AP. They work but I have the mother of all hums, likely ground loop induced which I will need to play with to fix.
In relation to that, where are people grounding their amps to the chassis? I have run a ground wire from the high current power ground input to chassis. Another alternative would be from the centre of the off board filter caps. I might need to experiment with lifting the board from chassis ground with a low ohm resistor too. Time to get out the clip leads.
Finally thanks to Owen for all the hardwork, and the great boards.
Rob.
In relation to that, where are people grounding their amps to the chassis? I have run a ground wire from the high current power ground input to chassis. Another alternative would be from the centre of the off board filter caps. I might need to experiment with lifting the board from chassis ground with a low ohm resistor too. Time to get out the clip leads.
Finally thanks to Owen for all the hardwork, and the great boards.
Rob.
Last edited:
No No No !!!
the connection from Main Audio Ground to Chassis may have to pass Fault Current if something really serious goes wrong with your mains powered equipment.
A resistor used for this connection potentially kills any one who touches any other component connected to that "lifted ground" amplifier.
This is a DEATH TRAP.
Do not ever put a resistor in a Fault Current escape route.
Read all the many Threads here, or go to ESP and read about it there.