I'm wanting to build a carlosfm type regulated supply but my transformer is +-35Vac which after rectification is giving rails of +-48Vdc. I want to regulate this down to about +-40Vdc but the LM338 regulators suggested in (Building a Gainclone chip amp with a regulated power supply (PSU).) have a max input voltage of 40V.
I see that LM317HVT are rated at 60V but have max currents of 3.7A whereas the LM338s have max currents of 5A. Will LM317HVT do the job? Are there others?
Thanks,
Simon.
I see that LM317HVT are rated at 60V but have max currents of 3.7A whereas the LM338s have max currents of 5A. Will LM317HVT do the job? Are there others?
Thanks,
Simon.
The LM338 is a floating voltage regulator. The regular version will tolerate a drop of 32 volts across it. You have to arrange the circuit so that this is not exceeded. Then you can regulate a few hundred volts if you want.
Have a look at the datasheet. If you can't figure it out from that, get back.
w
Have a look at the datasheet. If you can't figure it out from that, get back.
w
No 3-terminal adjustable reg that I'm aware of (inc LM338) has a max input voltage spec. That's because none has a ground connection. What they have is a max input-output voltage spec. So long as you don't short circuit or otherwise overload them, they'll work fine on higher voltages. An external voltage limiter can be employed so they don't blow up if they do get overloaded.
Thanks wakibacki.
The circuit I'm trying to follow is http://www.decdun.me.uk/gc/snub.reg.psu.png but with the voltage defining resistors R3-R4 set at 120R and 3.7k which I believe (hope) outputs 40V.
The application is for a gainclone using a TDA7293 which has a max voltage limit of +-50V so I'm trying to add a bit of safety margin with the regulator. Output of say 50W rms? but I'm not to picky.
Simon.
the 7A limit of the 3886 is already a bit too low for an 8ohm reactive load. I don't have TDA data for max current.
Neither a 1.5A nor a 5A current limited regulated supply is suitable for a real speaker load. Unless you can provide a short term source of transient power after the regulator to satisfy the demand of the power amplifier.
That short term power source can be very big capacitors just like used on an unregulated supply.
But read the datasheet for all the regulators you are considering. Many do not like a big capacitance on their output.
You will need to ensure the regulator you buy matches the datasheet that you have used for your design.
Neither a 1.5A nor a 5A current limited regulated supply is suitable for a real speaker load. Unless you can provide a short term source of transient power after the regulator to satisfy the demand of the power amplifier.
That short term power source can be very big capacitors just like used on an unregulated supply.
But read the datasheet for all the regulators you are considering. Many do not like a big capacitance on their output.
You will need to ensure the regulator you buy matches the datasheet that you have used for your design.
Last edited:
AndrewT - your expertise is beyond questioning, but what I do not understand is that many claim to have built regulated supplies for gainclone type amps and say they sound very good. So I'm confused when you say they wont work.
Simon.
chipamps do work.
I have listened to quite a few, without regulators.
But the 7A of the 3886 (which is amongst the highest available) is a serious limitation to excellent performance.
I cannot get Medium to High SPL from a chipamp powered 88dB speaker.
I don't believe that my Medium level listening is voltage clipping. I am more inclined to believe it is current clipping.
High level listening is very probably voltage and current clipping and it sounds terrible.
I have listened to quite a few, without regulators.
But the 7A of the 3886 (which is amongst the highest available) is a serious limitation to excellent performance.
I cannot get Medium to High SPL from a chipamp powered 88dB speaker.
I don't believe that my Medium level listening is voltage clipping. I am more inclined to believe it is current clipping.
High level listening is very probably voltage and current clipping and it sounds terrible.
So this is just a matter of belief what you're propagating about LM3886s? Why not measure and verify?
do you have the gear and expertise to do that for me?
But at least I have stated they are my beliefs.
It is up to the Members to decide to ignore my comments and/or find ways to corroborate rather than simply accept.
I expect the Members to make informed decisions, never to go in blind and be led by the blind.
But at least I have stated they are my beliefs.
It is up to the Members to decide to ignore my comments and/or find ways to corroborate rather than simply accept.
I expect the Members to make informed decisions, never to go in blind and be led by the blind.
Last edited:
OK, in a desperate attempt to regain control of my topic, might someone be able to advise me whether if I use the following circuit http://www.decdun.me.uk/gc/snub.reg.psu.png but with the voltage defining resistors R3-R4 set at 120R and 3.7k (which I believe/hope will output 40V.)
a) I wont fry the LM338s
b) I might have a power supply which might make a half decent job of driving a TDA7293 driving a 8 ohm load at ~50WRMS?
Ta,
Simon.
a) I wont fry the LM338s
b) I might have a power supply which might make a half decent job of driving a TDA7293 driving a 8 ohm load at ~50WRMS?
Ta,
Simon.
Power supply regulator
Hi,
Try using the TL783 as they recommended in their application manual. I used it feeding the TL783 output to the gate of an N-channel mosfet transistor gate "IRF640". You should be able to adjust the output from 1.25 volts to 120 volts. Do not forget to add a diode cathode connected to the source and the anode to the drain.
This is like having a variable zener diode.
Regards,
tauro0221
Hi,
Try using the TL783 as they recommended in their application manual. I used it feeding the TL783 output to the gate of an N-channel mosfet transistor gate "IRF640". You should be able to adjust the output from 1.25 volts to 120 volts. Do not forget to add a diode cathode connected to the source and the anode to the drain.
This is like having a variable zener diode.
Regards,
tauro0221
I think your voltage calculations are a little adrift because you didn't take account of your transformer regulation. Let's say you've got something like a 160VA toroid, regulation will be around 11% (value from Airlink Transformers). So the offload voltage will be considerably higher than your projected 48V, reaching around 54V after rectification. So I'm going to say you do stand a fairly high risk of frying them under overload conditions.
50W into 8R requires 28.2V peak. From memory I think the TDA will swing to around 3V of its rails, so you only need +/-32V. Your 35V transformer looks to have too high voltage for this purpose. You won't need regulators at all if you use a 25V transformer which will give you around 37V unloaded.
An alternative is to look at buck converters such as National "Simple Switchers" LM5088, LM5086 and LM5116 will all work with in your voltage and current needs. You need to add a Pi filter to filter out the noise. For very low output noise use a "coupled inductor" filter with can reduce noise below 1 to 5 mV.
Sorry, I got a bit distracted.
There's a lot to what AndrewT says, he's well into chipamps, and it's generally accepted that regulated power supplies are of dubious benefit with chipamps, precisely because mostly they don't cope well with a lot of capacitance after the regulator, and the amps generally have excellent PSRRs anyway.
abraxalito's point about the transformer regulation is good too. Although the calculations based on the nominal voltages give you ~48V this will probably rise to ~54V when lightly loaded.
This said, you still have the problem of how to make use of the transformer with your chosen chip.
The problem is that even with a +-48V supply (and a +-45V swing from the chip) the amp will be trying to supply ~125W into 8ohms. Suppose the amp is 50% efficient, this gives a peak current of ~11.25A. Which is a lot.
These numbers are for one channel only, too.
'Course you might say, I won't drive it that hard... but this is not really designing, more throw it together and hope.
You can regulate down to the ~32V that you need for 50W, but now you're dropping ~18-20V across the regulator, loaded. Which with the ~5A (RMS) that you need for 50W with 50% efficiency is a big 100W dissipation in a linear reg. Per channel, again.
The last solution from TechGuy is in many ways the most workable as switching regulators are comparatively efficient. They are less straightforward though.
The simplest solution would be to exchange the transformer. Or use it for something else and buy another.
Stew over these numbers for a while and see what you think. Try working up a spreadsheet to show you the RMS and peak currents and voltages, taking into account the amplifier efficiency, it's quite revealing. If you use a spreadsheet you can do a lot of what-if variations easily.
w
There's a lot to what AndrewT says, he's well into chipamps, and it's generally accepted that regulated power supplies are of dubious benefit with chipamps, precisely because mostly they don't cope well with a lot of capacitance after the regulator, and the amps generally have excellent PSRRs anyway.
abraxalito's point about the transformer regulation is good too. Although the calculations based on the nominal voltages give you ~48V this will probably rise to ~54V when lightly loaded.
This said, you still have the problem of how to make use of the transformer with your chosen chip.
The problem is that even with a +-48V supply (and a +-45V swing from the chip) the amp will be trying to supply ~125W into 8ohms. Suppose the amp is 50% efficient, this gives a peak current of ~11.25A. Which is a lot.
These numbers are for one channel only, too.
'Course you might say, I won't drive it that hard... but this is not really designing, more throw it together and hope.
You can regulate down to the ~32V that you need for 50W, but now you're dropping ~18-20V across the regulator, loaded. Which with the ~5A (RMS) that you need for 50W with 50% efficiency is a big 100W dissipation in a linear reg. Per channel, again.
The last solution from TechGuy is in many ways the most workable as switching regulators are comparatively efficient. They are less straightforward though.
The simplest solution would be to exchange the transformer. Or use it for something else and buy another.
Stew over these numbers for a while and see what you think. Try working up a spreadsheet to show you the RMS and peak currents and voltages, taking into account the amplifier efficiency, it's quite revealing. If you use a spreadsheet you can do a lot of what-if variations easily.
w
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.