current control (tricky one)

[Ubergeek63]

Ubergeek63,

a PWM power supply? I like the sound of it! More info please!

The chip is meant for current sourcing power LEDs so it is a natural for driving current into a joint.

The problem with "Just the old transformer for stepdown" is that at 1200W it will be huge. You need the "high" frequencies to get the size of the inductor down. It is imperative that you use switchmode since even a 50VA transformer will run $30 and up in that power range $200 plus shipping on something 8" diameter by 4" high at 25lbs.

https://www.electro-tech-online.com/custompdfs/2008/04/AN-H50.pdf is a good appnote on it, though what I suggest is going a bit further than that. And http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=M8374-ND would be a suitable inductor if a bit weak on the current.

It would run buck mode and I believe will reliably deliver 27A into 20V from a rectified mains. The formulas are all in the sheet. The other irritating detail is that this will NOT be isolated, which means that the entire circuit is tied to the mains. On the other hand, it IS the least expensive.

I would add that at the high levels in this discussion I would add an FET driver.

 

[bmachining]

Hmm, well i have a couple of transformers suitable already. I would really like it to be isolated from the mains. Been bitten by 240 a few times.
Am i correct in saying i need something like u have mentioned to drive a mosfet or 2 or more,which is fed from the transformer, with either built in or add on current sensing circuit to control the output via the pwm controller?

Then the Inductor and maybe some caps to smooth the output.?

 

[Ubergeek63]

Hmm, well i have a couple of transformers suitable already. I would really like it to be isolated from the mains. Been bitten by 240 a few times.
Am i correct in saying i need something like u have mentioned to drive a mosfet or 2 or more,which is fed from the transformer, with either built in or add on current sensing circuit to control the output via the pwm controller?

Then the Inductor and maybe some caps to smooth the output.?

in that case use one for isolation. The 1.5-2KVA isolators go for over $200 for the cheapies and cost a lot to ship.

Welders do not typically worry about clean DC. That circuit already has current sensing, it is a current regulator not a voltage regulator. If the transformer you have puts out 40VAC so much the better, a relatively simple tweak will approximate PFC as well. It will run CCM and I can show you how to make it reasonably PFC and not use big caps.

The $10 inductor, a couple $5 caps, an ultrafast 50A diode, and a 50+A FET (they tend to be pickier)

 

[bmachining]

PFC as in power factor correction? just read up on it so i have a very vauge idea now hehe. Transformer can put out any AC voltage i desire, rewound it myself, and not hard to change.

 

[Ubergeek63]

PFC as in power factor correction? just read up on it so i have a very vauge idea now hehe. Transformer can put out any AC voltage i desire, rewound it myself, and not hard to change.

Exactly. It is not the transformer or the rectifier that destroys the power factor but the filter cap.

With the current source chip you can use AC output of the transformer for your current source reference to the LD (Linear Dimming) pin and a pot to adjust the average current and be able to get almost the full VA from the transformer as power out to your load (I believe you are trying to do resistance welding)

 

[bmachining]

This will output DC or AC ? needs to be DC.
It will be used for electroplating. I have heard that fets amp rating will only be safe to about half value for continuos use ( constant), can these be paralled ?

 

[Ubergeek63]

pulsating DC. Electroplating makes sense, but I still do not think you need it fully filtered.

You generally want some headroom, but you do not need double the current capacity. It is more important that you have enough heat sinking. This one is reasonable and would handle it by itself: http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=497-5166-5-ND
Assuming you are using your step down.

 

[bmachining]

Unfortunatly it does need to be rather smooth, less than 5% ripple for some applications. does this complicate things?
My previous linear circuits are always well heatsinked, also with cooling fans.

Awesome work so far!! Thankyou for this help.

 

[Ubergeek63]

Unfortunatly it does need to be rather smooth, less than 5% ripple for some applications. does this complicate things?
My previous linear circuits are always well heatsinked, also with cooling fans.

Awesome work so far!! Thankyou for this help.

Not a problem...wont need nearly as much heat sinking.

I would not have figured it would need low ripple. There is a trade off in that case for what would cost less. a simple isolation transformer and high voltage parts or your 40VAC transformer with a big mother cap, this one might be a good trade off: http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=P10035-ND ... you need to keep the ripple valley over the output voltage, the current regulator will do the rest.

The ripple current in the cap will determine it's life and you still want the small $5 caps around it to take the high frequency current.

 

[kell]

What are you electroplating? 40 volts it way higher than you need for most stuff, like zinc only takes up less than one volt and other metals 2 or 3 volts.
So you don't want linear current regulation, you would burn up just about every bit of your available power simply in the regualtion. You could do it with SMPS techniques, but a simple buck regulator probably isn't the ticket for such a big voltage ratio. Are you up to high power PWM current source regulation? Why not use a transformer to lower the voltage and increase the current first, then worry about current regulation?

 

[bmachining]

Hi Kell, The supply will be for Hard Anodizing mainly, which takes serious voltage
due to the temperature the baths run at. I will be using a transformer.

Ubergeek, lowest cost isnt necessary. A few $$ on caps certainly isnt going to break the budget.

 

[kell]

bgmicro.com has 100uH 13 amp inductors for $1.50, which is way cheap:
**broken link removed**
You could parallel several in a buck circuit of the sort in the datasheet of the chip ubergeek mentioned and get your thirty amps.

 

[Ubergeek63]

bgmicro.com has 100uH 13 amp inductors for $1.50, which is way cheap:
**broken link removed**
You could parallel several in a buck circuit of the sort in the datasheet of the chip ubergeek mentioned and get your thirty amps.

Nice price...I don't know how long they stay there as BG does not look like the sort of place that has anything regular-all special purchases.

 

[Ubergeek63]

Hi Kell, The supply will be for Hard Anodizing mainly, which takes serious voltage
due to the temperature the baths run at. I will be using a transformer.

Ubergeek, lowest cost isnt necessary. A few $$ on caps certainly isnt going to break the budget.

Well that certainly makes it easier to find parts... you are looking at losing 60W in a 50A bridge, 50W in the freewheeling diode, and 50W in the FET.

The losses are one reason there are a lot of off-line switchers. Some of the parts cost more but the losses go down, at least in the bridge. The losses in the FET are harder to guess at, but the losses in the bridge go down to 10W in a 10A bridge at 240VAC input.

It is hard to say what a hand modded switching transformer would act like. There are batch variations even in machine wound ones!

 

[bmachining]

Nice link, thankyou!
Help me understand this a little better....
Is the Fet being supplied AC or DC? If AC then do we need a Bridge rectifier after it to get the DC?
If its DC then obviously the bridge is before the Fet.

Loosing some efficiency isnt a big issue, it usually goes in heat i imagine, can always compromise!

 

[Ubergeek63]

Nice link, thankyou!
Help me understand this a little better....
Is the Fet being supplied AC or DC? If AC then do we need a Bridge rectifier after it to get the DC?
If its DC then obviously the bridge is before the Fet.

Loosing some efficiency isnt a big issue, it usually goes in heat i imagine, can always compromise!

Transformer-bridge-BIG cap-cap-inductor-FET-diode-cap

The two little caps are for the high frequencies,although little is a relative term.

If you are actually putting out 1200W DC, the transformer and the line will see around 2KVA. That is why there are so many governments bowing to the utility companies demand for PFC. It is much easier to supply power more efficiently than it is to replace all the lines and transformers when they blow.

 

[Ubergeek63]

on the output side you probably want 100uF minimum at 50VDC and 30A RMS if you want real reliability, though 10A RMS is probably plenty. A similar value on the input, though the RMS is not as critical.

The higher the cap and inductor values the lower the ripple. If you are actually running CCM the cap RMS requirements are much lower. It would be more normal for that chip to be running CCM.

 

[bmachining]

Do i need a special type of diode? i did a search for 50A ones, but theres moe types than i realise.

Parts list so far,
HV9910BLG-G driver
STW120NF10 Mosfet
ECE-T1KP153FA Cap
100uH 13.2A Choke 3 of


a few smaller caps, resistors, diodes, pot, .......

Anything obvious missing?

 

[Ubergeek63]

the 9910 is a regulator...you need a mosfet driver as well

 

[bmachining]

High side or low side?