SSLV1.1 builds & fairy tales

Salas,
that "guide" value for 2A to 2.5A for a 40W device is totally misleading.

eg.
Take 2A and 10Vds resulting in Pq=20W
Rth s-a 2C/W.
Ta = 40°C
Ts ~ 40 + [2 * 20 * 1.2] ~ 88°C
Rth c-s ~ 1C/W
Tc ~ 88 + 1 * 20 ~ 108°C
De-rating factor = [150-108]/[150-25] = 42 / 125 = 0.336
The absolute max Pd for Tj =150°C is 13.44W

For a lower continuous Tj I would recommend a continuous Pq << 13.44W, maybe 8W.
That leaves you continuous current @ ~800mA for 10Vds.

Can you see why your 2A to 2.5A is totally ridiculous?

Now increase the mains voltage by 6% and recalculate the worst case Vds voltage. Is it 12V or 13V or 14V?

What would be the short term Tj for this continuous situation?
 
2C/W is hardly small.
I have used 5C/W and 7C/W and 10C/W in high current versions.
And I use 20C/W & 30C/W in ordinary DCB1.

And that 2C/W is only taking the one mosFET.
If all 4 mosFETs were fitted it would need to be bigger than 1C/W

Model your suggestion and let the Members see what you are proposing.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
2C/W is hardly small.
I have used 5C/W and 7C/W and 10C/W in high current versions.
And I use 20C/W & 30C/W in ordinary DCB1.

And that 2C/W is only taking the one mosFET.
If all 4 mosFETs were fitted it would need to be bigger than 1C/W

Model your suggestion and let the Members see what you are proposing.

DCB1 hot rod at 1A runs at 0.5C/W per side. Looks like an F5 amp cousin. I said 40W parts can be safe at half their rating on GOOD sinking, then you come up with YOUR 2C/W definition of good sinking for a 2A biased semi. Before you paint a general answer as a ridiculous statement, its better you ask the other party what he defines as good sinking for the job first. Now that would be a serious approach.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
@ wushuliu about thermal design.

Each time you want to know how big a sink you will need with a device gather 3 pieces of information first.
1. Ambient temperature. I.e. how hot your house's inside runs on average.
2. How many volts and amps the semi(s) gonna have across.
3. Its junction to case thermal resistance spec RthetaJC.

Rules of thumb.
1. Don't allow beyond 100C junction temperature although your semi may have 125C max or more, that's for conditions safety margin and longevity.
2. Play down the sink's spec by 30% to cater for heat spreading non ideally on sink, semis mount as hot spots.
3. If the sink will be in a box add 20C to room temp. For free or top/side out of box correctly oriented heat-sinks in still air just room temp will do.

Example.
L.A. can be hot in the summer but you won't sit down listening long with more than 27C in your living room given you turn on the air-con or you prefer a Gin&Tonic outdoors instead. That chunky nice 0.5C/W sink you have been eyeballing has arrived. Lets start. Rule of thumb 2 makes it a 0.65C one. Assume you wanna sink two Mosfets on a positive polarity Bib 1.1 with same characteristics for simplicity. Say IRF9530s running at 1A CCS 10Vin-Vout and 20V 0.5A spare current output. That will set them at 10W dissipation each. 20W total. Sink will rise at 20W*0.65C/W=13C above ambient since its common to both semis. When you will close the box although it has some ventilation holes the heat builds up inside it. Rule of thumb 3 says 27C+20C=47C. 47+13=60C on your sink. You got those nice Sil-Pads to insulate the semis. Alas, there is transfer resistance. About 1C/W. You could use best CPU grade thermal paste smeared thinly on sink to smooth it out and go 0.7W/C or so but most will avoid the mess. So your 10W each dissipating Mosfet case is at 60+10=70C for 1C/W case to sink. Info 3 when you will look up IRF9530 gives ~1.7C/W RthJC. Thus for 10W the hot nucleus of each Mosfet is 17C hotter than its case. 70+17=87C. Under rule of thumb 1 is good to go. Others may relax the rules but that's how I go with sinks for staying carefree in hot rod.
 
Let's look at your 2.5A @ 10Vds using first 0.5C/W and then 0.2C/W for a heatsink for a lone 40W mosFET.

Ta = 40°C, Rth s-a = 0.5C/W, Rth c-s = 1C/W, Rth j-c = 3.12C/W
Pq = 25W continuous.
Delta Ts-a = 25 * 0.5 * DFsink = 12.5 * DF, I'll assume DF for a 12.5C delta is 1.5.
Ts = 40 + 12.5 * 1.5 = 58.75°C
Delta Tc-s = 25 * 1 = 25C
Tc = 58.75 + 25 = 83.75°C
Delta j-c = 25 * 3.12 = 78C
Tj = 83.75 + 78 = 161.75°C > 150°C, outside manufacturer's specification.
We have not allowed for a rise in mains voltage yet. I'll do that when I consider the 0.2C/W sink.
DFsoa = [150 - 83.75] / [150 - 25] = 0.53.
This tells us that the 40W device can dissipate 21.2W when Tc is @ 83.75°C

Now to the big sink sitting outside the chassis.
Ta = 30°C, Rth s-a = 0.2C/W, Rth c-s = 1C/W, Rth j-c = 3.12C/W
Pq = 25W continuous.
Delta Ts-a = 25 * 0.2 * DFsink = 5C * DF, I'll keep the same assumption DF for a 5C delta @ 1.5.
Ts = 30 + 5 * 1.5 = 37.5°C
Delta Tc-s = 25 * 1 = 25C
Tc = 37.5 + 25 = 62.5°C
Delta j-c = 25 * 3.12 = 78C
Tj = 62.5 + 78 = 140.5 < 150°C, within manufacturer's specification.
Now look at the mains voltage variation.
at nominal voltage the mosFET sees 10Vds.
Vout = 10Vdc, Vdrop = 10Vds, Vin = 20Vdc+[half of ripple], Vat rectifier = 20.7Vdc+{half of ripple] I'll guess that Vripple =500mV @ 2.5A of output
Vac at transformer = 20.95 / sqrt(2) = 14.8Vac

Now increase the mains by 6%. The Vac rises to 15.7Vac.
Vdc = sqrt(2)*15.7 - half ripple - diode drop = 21,2Vdc
This gives Vdrop = 11.2Vds
Pq = 28W continuous but relatively short term, maybe an hour or 30 minutes.
Tj = 30 + 28 * [0.2*DF + 1 + 3.12] ~ 153.7°C > 150°C
We still have to short the output to check if the mosFET can survive during this hopefully short term worst case condition.

Can anyone see why the original suggestion of >= approximately half Pdmax is a nonsense?

If advice and guidance is being offered, it seems to me that that advice should be based on fact or experiment.
The advice must not be based on some plucked out of the air number that happens in this case to NOT be suitable.

If there are any mistakes in my arithmetic, or a formula I have wrong, then please inform and I will correct the script. I want the information and the method of modeling to be right, for all those who see a need to simulate the actual operating conditions of their CCS mosFET.
Similarly, if you see a need for a bit more explanation within the script, then make suggestions.
 
IRF9530s running at 1A CCS 10Vin-Vout and 20V 0.5A spare current output. That will set them at 10W dissipation each. .........
this 10W for Pq cf. 88W device (~12% of Pdmax) is much more sensible and the sim shows a long term Tj of <90°C.
No comparison to the earlier misguided advice of 50% to 62%.

Salas,
you have shown how to do it (model or simulate) rather than simply guess. I knew you could do it. What I can't understand is why you offered that 2A to 2.5A guidance in the first place and then tried to defend the situation when it was clearly wrong.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
I did not offer any sinking guidance. Just commented on amperage pictured on a SOA chart. Someone could use a fan or a water cooler or stick it in the fridge. Nobody asked on how to sink that amperage on what sink he has got in hand. Your style, assumptions, and attitude are not appreciated. Don't address me in my threads again unless you want to be ignored.
 
Dear Salas,

Thank you for your generous contribution of this marvelous power supply. Also very grateful to everyone for their posts in this thread too, it's been helpful to read all your comments. I'm about to start on my +5V, 150 mA power supply too. I think someone built a PS for his ODAC with the same specs as well. I just want to confirm a few things, please bear with me :eek:

1. Can I use 50VA 2x22V for my 5V out target? Or is it too much?
For 22VDC in, Excel calculator gives me estimated minimum heatsink rating of 3.27 deg/W. Is this a reliable figure? Should I run out to get a lower spec’ed 12-15V transformer so I don’t have to dissipate so much heat on Q301?

2. From excel calculations, R301 = 12.2 ohm, 3W (calculated 1.1W). Where do I buy an exact 12.2 Ohm resistor?

3. Is R303 = 220R (?) Referencing what Salas said earlier, “You need the 220R or a simple diode in series with the LEDs. You need to jumper the non used positions to close the series Vref circuit. Use a 33R dummy to simulate 150mA load.”

If 3. is correct, then to double confirm, for the Norton section, R303 = 220R, R305 = 1k Trimmer, 1x Red LED on EITHER D305 OR D306, and jumper EVERYTHING else in the Norton section?

4. R305 = 1K Bourns 3296 Y trimmer (Read people mentioning about replacing this with a quality, fixed resistor once trimmed resistance to target. How do I find out what resistance the trimmer’s at? Do I need to get z201/mk132 type or any will suffice?)

5. Is C302 and C304 = 220 uF preferred for digital? Or C302 = 220 uF and C304 = MKT; R307 = 1R better?

6. For C305, is 10,000 uF of this Panasonic alu cap model ok? I can’t get the 10,000 uF panasonic FC, nichicon muse and elna silmic types. Or should I get 4700 uF FC/muse/silmic?

Thanks in advance for all the help :)
 
diyAudio Chief Moderator
Joined 2002
Paid Member
1. Use a proper secondary transformer 12-15 Vac.
2. Is 150mA your current source setting or your load consumption? Be sure you use more CCS than the load needs. Closest to target available R301 resistor values are fine. Leds and Mosfet will have tolerances too. The CCS setting in this application is just to cover the plan, no exactness plays any role.
3. Use a diode in same orientation as the ''see text'' PCB area Leds. It will give you a voltage drop that will be self set, when with a resistor it will depend enough on Q303's Idss that energizes the Leds. Its either you use 1k trimmer and one red led, or 2 red leds and a diode for 5V. The diode position can alternatively be a ~220R for easier filtration by C301 but it may give you hassle trimming it is the deal. Its a small chance it will be grossly off target with a 220R though. Don't go after 5V dead on, go after you are in your client circuit's acceptable power voltage area.
4. You better take it out of PCB after a setting is good and you need measure the set value to sub a resistor for. Measure from wiper to the leg it was installed towards R304. Get a Dale low ppm fixed resistor or analogous to upgrade from a trimmer.
5. 220-1000uF C301 are OK values for digital, MKT & 1R termination are OK.
6. Looks like a plain type Nichicon 1000uF snap-in that in the link. Get a 4700uF FC or Muse, it fits more than well for just 150mA.
 
1. Okay, will this transformer do? I'm on 230v. Any advantages/disadvantages you personally found with r core vs toroid vs ei?
2. 150 mA is load consumption. Okay, 300 mA CCS good to go?
3. "Its either you use 1k trimmer and one red led, or 2 red leds and a diode for 5V"
If i'm reading you right, 2 red leds + diode (D301), everything else jumpered, no need for trimmer in R305 and no need for R303?
"The diode position can alternatively be a ~220R for easier filtration by C301" means instead of diode in D301, alternatively it can be ~ 220R?
4. Okay, does the quality of resistor matter much here?
5. Will 1000 uF be better than 220 uF for C301?
6. For 300 mA CCS, will 4700 uF be ok?

Thanks a million, Salas!