Power supply simulator. / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Power supply simulator.

previous 1 2 3 4 next last

Move Thread

LAN_403

DerAlbi

Fri May 04 2018, 05:44PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

The inevitable question must arise now: if all this is necessary, is Class-A the right topology? There seems to be a concept problem, when the only counter measure to the resulting problems is along the lines of "as much as possible"

If its for fun and money burning, then its ok of course.. ^_^

Ash Small

Sun May 06 2018, 02:46PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

DerAlbi wrote ...

The inevitable question must arise now: if all this is necessary, is Class-A the right topology? There seems to be a concept problem, when the only counter measure to the resulting problems is along the lines of "as much as possible"

If its for fun and money burning, then its ok of course.. ^_^

The purpose of the exercise is to build a single ended Class A amp using the new SiC power JFET's from United Silicon Carbide. I managed to get hold of five of the 21A 1200V ones. but they are currently unavailable in Europe, as far as I'm aware. Not sure if they are currently available in US. I've as yet been unable to get hold of any of the 38A devices.

I'm currently considering using chokes in a parafeed type arrangment, and ordered a few extra cores to experiment with

2Spoons

Sun May 06 2018, 09:56PM

Registered Member #2939 Joined: Fri Jun 25 2010, 04:25AM
Location:
Posts: 615

I have to agree with DerAlbi: this is starting to look like one of those situations where you've decided to use glue, but now you are needing so much glue that in hindsight maybe a bolt would have been a better idea.

So, rather than getting too crazy with supply filtering perhaps its time to look at improving the PSRR of your amplifier? Or maybe look at using a switching reg followed by a more sensibly sized LC filter?

Sulaiman

Sun May 06 2018, 10:46PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

It just occurred to me - how will you heatsink the transistor ?
I've recently been playing with the eBay GPU/CPU water cooling blocks, TECs and old CPU heatsink/fan combinations,
In this case TECs would be too low power so you may need a source of chilled running water,
or a massive conventional forced-air heatsink.
========================================
Also, transformer output Class-A amplifiers are usually very tolerant of power supply ripple

hen918

Sun May 06 2018, 11:36PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

I think that the stage has already been well passed where any improvement could have been detectable.

Ash Small

Mon May 07 2018, 01:24AM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Sulaiman wrote ...

It just occurred to me - how will you heatsink the transistor ?
I've recently been playing with the eBay GPU/CPU water cooling blocks, TECs and old CPU heatsink/fan combinations,
In this case TECs would be too low power so you may need a source of chilled running water,
or a massive conventional forced-air heatsink.
========================================
Also, transformer output Class-A amplifiers are usually very tolerant of power supply ripple

I've five 21 Amp transistors (JFET's), I figure if they are good for 21 Amps continuous, they shouldn't be too hard to keep cool at 1 Amp each (1 Amp each is the maximum, they will probably run on less)

There will be another half a dozen or so stages running a few hundred milliamps between them, input stage, pre-amp stage, active tone control, reverb tank drive and recovery, and maybe one or two more. This still leaves at least four Amps for the power stage, if I need it.

There is no output transformer.....Output impedence is so low I expect to have to parallel 8 Ohm speakers to match the impedence. It just has a huge output capacitor, maybe 1000 uF or more......For a 1 Ohm load impedence I've calculated 3000 uF, but I'm now thinking four Ohms may be plenty, as this power supply delivers more volts than other alternatives. Zeners, regulators and resistors all drop volts.

I've been winding a choke this evening, I'm up to 99 turns and 1.7mH. I think another twenty turns or so should get me to around 2.2mH.

DerAlbi

Mon May 07 2018, 02:25AM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

I've five 21 Amp transistors (JFET's), I figure if they are good for 21 Amps continuous, they shouldn't be too hard to keep cool at 1 Amp each

Thats an extremely bad argument (and i think that you know better) but i cant let it stand there if anyone else picks up on such a thought pattern.
The heat capability of a mosfet has nothing to do with its current capability! If the Mosfet is rated at 21A and 1200V does not mean that both conditions may apply at the same time (which equates to 25.2kW on the device which is obviously ridicules)
For such considerations the datasheet gives you a Safe Operating Area (SOA-diagram). This is shows the user of the device which conditions are permitted. It is extremely important for Class-A operation that in such a diagram is a DC-curve shown, if not the device is not specified for linear operation and is made for switching only. Another catch on the SOA diagram is that it plainly lies since it assumes perfect cooling on the device to 25Â°C. Any permissible heat output shown in the SOA must be derated according to actual operating temperature.

Ash Small

Mon May 07 2018, 04:00AM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

DerAlbi wrote ...

I've five 21 Amp transistors (JFET's), I figure if they are good for 21 Amps continuous, they shouldn't be too hard to keep cool at 1 Amp each

Yes.

I'm aware of that. My plan was to run them up slowly (limit the current), then increase the current slowly until they get hot. Once they are too hot to touch, they are too hot

There are no SOA curves for them, they are designed for switching, but I've had very good results with other JFET's designed for switching. I'm using J105's in the pre-amp stages, no-one else seems to use those for audio.....yet.....

This is one of the reasons I'm trying to keep the voltage as high as possible, and if I do run less current, voltage gets a bit higher. At present I'm planning to use some of the voltage for biasing the JFET's, although I may consider a separate B- supply, it doesn't have to supply much current, it will be loaded with a megohm or so.

The design is still very experimental, I've not really finalised anything yet, I thought it best to start with a decent regulated power supply. the simulator gives a ripple figure measured in uV, and if I add more inductance I should be able to clamp the current to some extent, which should be of benefit.

I've not done any calculations on the amount of inductance I'll need to make a worthwhile improvement, but it should be a few orders less than the choke needed for a tube amp running milliOhms.

Like I said earlier, it's all very experimental, but based on 100 year old technology

Sulaiman

Mon May 07 2018, 10:07AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Ash, I guess that you do not understand impedance matching at the output
"There is no output transformer.....Output impedence is so low I expect to have to parallel 8 Ohm speakers to match the impedence."

A couple of examples;

Supply = 110 Vac full-wave rectified = 150 Vdc, output voltage swing = +/- 140 V
Quiescent (resting) current = 1.1 Adc, current swing = +/- 1A.
Output impedance = 140/1 = 140 Ohm.
For an 8 Ohm load a voltage transformation of SQRT(140/8) = 4.18
So the output transformer should have a pri:sec turns ratio of about 4.18 is required
The transistor will dissipate 150V x 1.1A = 165W ... a significant heatsink required
Output power = 70 Wrms

Supply = 240 Vac = 320 Vdc, output voltage swing = +/- 300 V
Quiescent current = 550 mA, output swing = +/- 500 mA
output impedance = 300/0.5 = 600 Ohm
Transformer ratio = SQRT(600/8) = 8.66:1
Transistor dissipation = 320 x .55 = 176 W
Output power = 75 Wrms

Supply = 320 V, Quiescent current = 550 mA, power dissipation = 176 Watts
Current swing = +/- 500 mA
IF THERE IS NO OUTPUT TRANSFORMER
Output = +/- 500 mA into 8 Ohms = 1 Watt rms
... not very efficient ;)
=======================================
As you will be allowing the semiconductor to get hot,
the biasing (quiescent current) be unstable because Vgs for constant Id varies with temperature.
(check datasheet)
I have only one SiC transistor that I've used for switching so I'm not sure about linear operation parameters.
===========================
You would have discovered these things as you experiment,
I've learned 'the hard way' so I hope the above helps.

You may notice that I assumed the output voltage swings to be less than the supply voltage,
this is because transistors become non-linear as the voltage across them approaches zero.
Similarly, the output current swing I used is about 10% less than the quiescent current,
as transistor operation becomes non-linear as current approaches zero.

For Class-A, the theoretical maximum efficiency (power out / power in) is 50%
In practice 45% is very good, with 33% overall efficiency not uncommon.
======================================== =======
Last one;
Suppose you go for the 320 Vdc 550 mA class-A amplifier,
the drain/collector output impedance was calculated to be 600 Ohms,
so for a frequency response 3 dB down at 20 Hz you need
transformer primary inductance impedance >= 600 Ohms @ 20 Hz,
Inductance >= 600/(2.pi.20) >= 4.77 Henry
The energy stored in the transformer primary inductance is 0.5 x L x I^2 = 0.72 Joules,
that will require a significant airgap in the core.
The same applies if just an inductor is used in the drain/collector to +Vdd/+Vss
So, just like in valve-amp days, the output transformer will be neither trivial nor cheap.

DerAlbi

Mon May 07 2018, 11:24AM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

they are designed for switching, but I've had very good results with other JFET's designed for switching

Again the thought pattern cant go uncommented: any device is different from another. Anecdotal evidence that something works out of spec [must be assumed if spec is not known] does not turn a good experience into good engineering nor builds it a valid experience for future device abuses especially if you change out the device and draw parallels where are none.

There are no SOA curves for them

Since you gave some details like "UnitedSiC", 21A, 1.2kV... and i really couldnt believe such statement [SOA is a way too fundamental spec] Mouser has a datasheet.

On page 6 bottom right, you will find the SOA.
Have fun at 100V @ 1A @ DC @ 25Â°C and turn the current down to 500mA when the case [= backplate metal] temperature reaches 100Â°C because then only 60W dissipation is allowed as seen in page 6 bottom left.

Once they are too hot to touch, they are too hot

No. Human temperature sensing has nothing to do with semiconductor limits. The threshold you describe is about 50Â°C - unless you can show in the datasheet which specific device property derates at this point so that your design wont work, the temperature is fine. You can run with no issues at 120Â°C an more as long as you respect the specification. Usually the 150Â°C - 175Â°C limit arises from the plastic packaging which puts stress on the device due to different thermal expansion compared to the copper backplate. If you are careful within specs you can accidentally run a device until it desolders itself [during prototyping, and technically thats not in spec..]

Have fun building!

previous 1 2 3 4 next last

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob