I guess that´s the only way to find out for sure!
A few thoughts: During warmup the FET will need to dissipate 70-100W for a few seconds so using two or more high power devices in parallel might be clever.
I´m a bit troubled by the lack of a negative voltage rail for the operational amps, would anything be improved by adding one?
A few thoughts: During warmup the FET will need to dissipate 70-100W for a few seconds so using two or more high power devices in parallel might be clever.
I´m a bit troubled by the lack of a negative voltage rail for the operational amps, would anything be improved by adding one?
I'm building a pair of these for 300Bs, and also looking at building
a version for Eimac 4-65A and 3C24s, admittedly smaller than what
you need, but still beyond the capability of this circuit as is.
The first thing that occur to me is that with 10A current everything
will need beefing up. The rectifiers will dissipate 4W and will need
heatsinking.
The MOSFET needs to be something like this that will operate with
approximately 0.5 volts drain to source at 10A current (5W diss.)
http://www.irf.com/product-info/datasheets/data/irfp3206pbf.pdf
The opamp is speced for rail to rail inputs and output swing from
0V to 1.5 V below V+, plenty to drive the MOSFET into conduction.
Circuit traces and connections also need beefing up for 10A current.
I built a single-sided PCB for a power supply once that had diodes
soldered in through the holes. After 24 hours of use, the heat from
the diodes had melted the solder and charred the board. The
customer noticed the front panel lights flickering and a funny smell...
Cheers,
Michael
Oh and of course R4 should be 0.01 ohms 5 watts
a version for Eimac 4-65A and 3C24s, admittedly smaller than what
you need, but still beyond the capability of this circuit as is.
The first thing that occur to me is that with 10A current everything
will need beefing up. The rectifiers will dissipate 4W and will need
heatsinking.
The MOSFET needs to be something like this that will operate with
approximately 0.5 volts drain to source at 10A current (5W diss.)
http://www.irf.com/product-info/datasheets/data/irfp3206pbf.pdf
The opamp is speced for rail to rail inputs and output swing from
0V to 1.5 V below V+, plenty to drive the MOSFET into conduction.
Circuit traces and connections also need beefing up for 10A current.
I built a single-sided PCB for a power supply once that had diodes
soldered in through the holes. After 24 hours of use, the heat from
the diodes had melted the solder and charred the board. The
customer noticed the front panel lights flickering and a funny smell...
Cheers,
Michael
Oh and of course R4 should be 0.01 ohms 5 watts
Thanks for your reply!
Regarding the fets I have a few high power, high current types (IRFP260, 75645P and others). As i mentioned earlier I´m planning to use two or more Fets per channel to spread the heat and increase their chances to survive the warmup procedure.
The PCB layout is of course useless in this application so my plan is to hardwire the whole thing on a pair of huge heatsinks using a tiny piece of veroboard for the IC. I have some 35A bridge rectifiers, a 2x15V 300VA toroidal transformer and some big caps and chokes to provide a fairly clean 16-18VDC for the active circuitry to work with.
I have some experience of building class A transistor amps so cooling mosfets and building high current power supplies is not new to me. Most of my concerns were regarding the single rail operation of the IC.
Regarding the fets I have a few high power, high current types (IRFP260, 75645P and others). As i mentioned earlier I´m planning to use two or more Fets per channel to spread the heat and increase their chances to survive the warmup procedure.
The PCB layout is of course useless in this application so my plan is to hardwire the whole thing on a pair of huge heatsinks using a tiny piece of veroboard for the IC. I have some 35A bridge rectifiers, a 2x15V 300VA toroidal transformer and some big caps and chokes to provide a fairly clean 16-18VDC for the active circuitry to work with.
I have some experience of building class A transistor amps so cooling mosfets and building high current power supplies is not new to me. Most of my concerns were regarding the single rail operation of the IC.
Sorry, I'm sure you know what you're doing.
I guess my point, besides the opamp thing, was that this VCCS does
not need much voltage to work with. It should be possible to do what
you need with 7-8 volts and low DCR components. You can avoid
dissipating another 100W in the filament supply itself.
Cheers,
Michael
I guess my point, besides the opamp thing, was that this VCCS does
not need much voltage to work with. It should be possible to do what
you need with 7-8 volts and low DCR components. You can avoid
dissipating another 100W in the filament supply itself.
Cheers,
Michael
Might be difficult to feed a 10V 5A filament with 7-8V, I guess you misread the numbers in my first post
Regarding voltages and transformers I have the following options:
1: use the 2x15V transformer and prepare to get rid of serious amounts of heat. I can do that but the transformer has a 220V primary and I´m not sure it can handle 230V mains without mechanical buzzing.
2: use a pair of 11,5V 160VA EI-transformers that I know can deliver 5ADC without making noises. I guess these would give about 13VDC after rectification and CLC-filtering. Enough for the circuit to work with?
3: make a deal with a friend and trade some tube stuff against a pair of 2x12V 150VA toroids and some 100kµF 20V caps.
(This option came up late last night).
hey-Hey!!!,
I'd feed your unregulated voltage rail from an LC filter. YOu should be able to start that at a minimal amount over the minimum headroom requirements. I'll just note that my 813 are run PP with individual TX's and hum pots and have no hum... SE is rather troublesome.
cheers,
Douglas
I'd feed your unregulated voltage rail from an LC filter. YOu should be able to start that at a minimal amount over the minimum headroom requirements. I'll just note that my 813 are run PP with individual TX's and hum pots and have no hum...
SE is rather troublesome.cheers,
Douglas
Fuling said:
Might be difficult to feed a 10V 5A filament with 7-8V, I guess you misread the numbers in my first post
Regarding voltages and transformers I have the following options:
1: use the 2x15V transformer and prepare to get rid of serious amounts of heat. I can do that but the transformer has a 220V primary and I´m not sure it can handle 230V mains without mechanical buzzing.
2: use a pair of 11,5V 160VA EI-transformers that I know can deliver 5ADC without making noises. I guess these would give about 13VDC after rectification and CLC-filtering. Enough for the circuit to work with?
3: make a deal with a friend and trade some tube stuff against a pair of 2x12V 150VA toroids and some 100kµF 20V caps.
(This option came up late last night).
Oops sorry, sure, I misread both your post and the 813 data sheet.
I read 5V 10A! It's really 10V 5A which is easier. Never mind ;-)
Still, the regulator should operate with less than 2V drop-out, so
you might be able to get away with the big low-DCR transformer
in the 12V range. It might still be good to reduce R4 to .05R and
use a MOSFET with really low Rdson, like 50mOhms or less. I guess
you'll need some headroom for line voltage variations.
Sorry about the confusion, I've cleaned my glasses and screen ;-)
Douglas, is that a critical-L input you're recommending? That would
need more on the secondary than 12V RMS, right?
Michael
Fuling said:Hi!
LC (choke input) filters has crossed my mind but the chokes would be true backbreakers...
I don't think they'd be too big. 20 mHy at 6A ought to cover it, and that is less than 10 pounds. Get much bigger and mass starts to get worrysome. I'm also wary of the charging spikes a cap input filter delivering 5A of DC would require...the power iron will have to be bigger if you have no choke. If I had to guess, root2 more weight for the power and choke v. the cap-input power TX.
Mike, the starting point is ~0,9x the RMS input AC. 12V would not do it; diode drop, and Ohmic would probably take more than 2V. We'd still want a lot of capacitance in order to tolerate and smooth the ripple current.
cheers,
Douglas
Michael: Yes, 10V and 5A it is. Way easier than 5V and 10A but still somewhat problematic, this amp would have been finishead years ago if it weren´t for the filament power supplies...
I have already prototyped a passive circuit with CLCLC-filtering using 1,1mH iron EI chokes and several 15k and 27kuF caps. Got the resicual AC level down to 3mV peak to peak, not sure if that´s good enough.
Active regulation seems like a more clever way to go (less bulky) and this voltage monitored CCS feels even more so.
I have some IRFP055 (?) FETs somewhere that I believe has extremly low Rdson, will check them out asap.
I also have a few 12V (+-10%) 100W SMPS units but I´m somewhat sceptical to them after observing the output voltage on my scope.
Regarding LC-filters I think something like 2mH 20A would be better suited for the input chokes.
AFAIK the critical inductance is 1mH/ohm and the choke should be rated 4x the current.
Hammond makes such chokes but I´m sure they don´t come cheap.
I have already prototyped a passive circuit with CLCLC-filtering using 1,1mH iron EI chokes and several 15k and 27kuF caps. Got the resicual AC level down to 3mV peak to peak, not sure if that´s good enough.
Active regulation seems like a more clever way to go (less bulky) and this voltage monitored CCS feels even more so.
I have some IRFP055 (?) FETs somewhere that I believe has extremly low Rdson, will check them out asap.
I also have a few 12V (+-10%) 100W SMPS units but I´m somewhat sceptical to them after observing the output voltage on my scope.
Regarding LC-filters I think something like 2mH 20A would be better suited for the input chokes.
AFAIK the critical inductance is 1mH/ohm and the choke should be rated 4x the current.
Hammond makes such chokes but I´m sure they don´t come cheap.
I always wanted to try one of these chips driving an external FET with a very low on resistance.
http://www.micrel.com/page.do?page=/product-info/products/mic5156.shtml
http://www.micrel.com/page.do?page=/product-info/products/mic5156.shtml
Playing around in PSUD, it looks like you get full conduction at 2-3
mH on the input choke and could use up to 5-10mH, but the DCR
needs to be minimized, 0.2 ohms is too high, 0.1 or even 0.05 is
better.
It looks like 15V RMS is needed with L input to get 12V DC ,
assuming reasonable DCR in the path. The 4X rating for Lcrit may
be more than you need for peak charging current, but desired for
the low DCR.
I'm making the same tradeoffs for the puny 18-20W filaments of
my 3C24 -> 4-65A PP, and I'm now leaning toward L-input, even
with the higher heat loss...
Thanks!
Michael
mH on the input choke and could use up to 5-10mH, but the DCR
needs to be minimized, 0.2 ohms is too high, 0.1 or even 0.05 is
better.
It looks like 15V RMS is needed with L input to get 12V DC ,
assuming reasonable DCR in the path. The 4X rating for Lcrit may
be more than you need for peak charging current, but desired for
the low DCR.
I'm making the same tradeoffs for the puny 18-20W filaments of
my 3C24 -> 4-65A PP, and I'm now leaning toward L-input, even
with the higher heat loss...
Thanks!
Michael
astouffer said:
Interesting.. The current limit could be a plus, as the posted circuit
is not short-circuit proof. The voltage flag output could be used to
control B+ when the filaments have reached 90% voltage also.
Definitely worth some more study to see if it can be used in a VCCS
circuit.
Thanks!
Michael
The chokes that I have are rated 10ADC and I have used them in LCLC-filtered filament supplies up to 4A (811A tubes). With two in series at the forst position they were just under the limit of what they could put up with without buzzing. Sadly I don´t think there´s a chance they would work as inputchokes in a 5A supply.
Using LC-filtering would of course be much easier on the caps as they wouldn´t suffer from heavy ripple currents. On the other hand I have some hardcore Rifa caps that are made for this kind of job...
Using LC-filtering would of course be much easier on the caps as they wouldn´t suffer from heavy ripple currents. On the other hand I have some hardcore Rifa caps that are made for this kind of job...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.