yeah I know bout that scene I usually just use a big cement resistor on the primary and a relay sensing the secondary voltage. works well but they burn up under some rare faults.
Can you set R15 to 200 % of its value? ;-) Otherwise how will it make 600 mV from 400 mV?
I suggest you to add 500 uA constant current to CL node. Or a voltage offset with a schottky diode.
That would defeat the challenge I've given myself to create the best linear possible based on the LM723 wouldn't it.
Best? If you don't expect such parameters as efficiency and speed of control... If your goal includes using LM723 than you must use it, yes.
Because...?
Do you have any equation?
Sorry, but I have very limited resources for this. No access to my office PC, limited free time, and no knowledge about what you know already. I cannot provide an overall equation for thermal system engineering.
Then I don't need to tell anything. It's good, be happy with it!
Have you done this?
For the fan, I dont have (I wont search for it), especially if you limit the noise. I just informed you that there will be not that much air flow by far as you stated when you place fan inside the box. But placement can be optimised. If you place fan inside the box, how it will cool the heat sinks outside?
Does it exist? Not BC549B? Whats the dissipation limit?
In this case oversizing is much easier than optimisation.
Youre welcome!
Hi FJ
Your schematic has many things that are worse off than an Astron. The fact that you limit yourself to the uA723 and then threw away their basic design, is starting off severely crippled, about the only good thing on that chip is the V reference specs. (I'd hate to be around when you go to test current limit and overvoltage.)
The most prudent thing for you (E.g. single chip design) is to copy Astrons basic design topology and beef up all areas of cheapness, E.g. chassis , components, and construction techniques. They know how to build stuff well, but sadly they pinched pennies everywhere to within inches of its ratings. Simply upgrade where it does the most good. E.g. their custom transformer is undersized, do use premium devices ( good industrial designs never use TO-3 sockets they are unreliable ) Do use slow start takes the strain off the rectifiers. Since you are familiar with repairing them & also much data exists from the field on what needs to be looked after, IMO that is an ideal situation. Since you are starting from scratch using that single chip approach there are many pitfalls & lessons to be learned ahead on your journey. good thing electronics is a fun hobby.
Your schematic has many things that are worse off than an Astron. The fact that you limit yourself to the uA723 and then threw away their basic design, is starting off severely crippled, about the only good thing on that chip is the V reference specs. (I'd hate to be around when you go to test current limit and overvoltage.)
The most prudent thing for you (E.g. single chip design) is to copy Astrons basic design topology and beef up all areas of cheapness, E.g. chassis , components, and construction techniques. They know how to build stuff well, but sadly they pinched pennies everywhere to within inches of its ratings. Simply upgrade where it does the most good. E.g. their custom transformer is undersized, do use premium devices ( good industrial designs never use TO-3 sockets they are unreliable ) Do use slow start takes the strain off the rectifiers. Since you are familiar with repairing them & also much data exists from the field on what needs to be looked after, IMO that is an ideal situation. Since you are starting from scratch using that single chip approach there are many pitfalls & lessons to be learned ahead on your journey. good thing electronics is a fun hobby.
Why do you feel that the 723 precision linear voltage regulator is a limiting factor? I have not found a more feature-packed, thoroughly tested, safe linear regulator available anywhere. Again, please point me to it if in fact one does exist. I've lost count of the articles and white papers I've read that state that there is really no great replacement in the linear realm for the LM723.
Other than the common-side Fuse which I'm removing from the design, why do you feel this way. For example, and as per the LM723 design specs, I'm using a zener voltage reference to trigger a crowbar scenario either directly or through SCR1 (as I stated that part of the design is not complete, but the circuit is in place even at this early stage on the schematic for the most part). So, what part of that reference standard design makes you worried? I'm sincerely interested in knowing if you don't mind sharing.
Other than my design not using a separate winding or supply for the regulation circuit, what do you feel I have deviated from that has led to a deficiency in my design? Although I am exploring the separate power supply recommendation seriously. My design has a better power factor, better chassis, better components, better cooling, thermal shut-down, remote voltage sensing, fallback current limiting, short circuit protection, variable voltage, better filtering, better metering (accurate), better rectification, and convenient status LEDs all in a simple circuit design which is easy to build and repair.
I agree that they can be problematic. However, I'm using some of the best sockets available on the planet to significantly reduce any potential pitfalls. They are made by TE Connectivity and cost $22 each at Mouser. That's $244 just for the TO-3 sockets on this build. I've never seen a typical Linear Supply which uses the TO-3 package not use sockets actually, including Astron. Unfortunately they use sub-standard sockets in their builds which lends to the problems of reliability you've stated here.
I found a few options on ebay, but they are cheap and unlikely to handle anything over 800 watts from what I see. I've reached out to a design engineer to assist with the crow-bar and soft-start aspects of this design.
Thanks for your time; much appreciated.
Great point; thanks I'll make the change to R15 right away. You also asked about why I've chosen the cooling Fan specifically. Because it is capable of 2.9 cubit meters of air flow, with a static coefficient of 67Pa and 102 CFM. Given those operating characteristics, and that the chassis (empty) has a volume of .02 cubic meters. Internal cooling will be the least of my concerns for this build. Additionally, the reason the internal Fan is a viable cooling option is because 100% of the left and right panels of the chassis are the finned heat sinks. So the airflow will be in direct contact with the massive cooling heat sinks 100% of the time during operation. Additionally, all of the pass transistors will be directly mounted to the heat sinks via aluminum brackets which place them directly underneath the vent slots in the top of the chassis, thereby creating a direct forced-air cooling environment. I'm not aware of any commercially available linear supply with these specifications that has better cooling design than this. And certainly not at 39dBA.
I had a typo, the correct part number is BC549BTA. The Pc is 500mW.
I'm using Tina to simulate the design, but it's frustrating because I can't find a viable LM723 (or ua723) Spice model. With my limited knowledge of how to manipulate simulation model characteristics in software, this is severely limiting my ability to properly simulate this design.
Thanks again for your time and especially for catching my math error for R15.
see 'Power One' for typical mounting of TO-3 used industry wide and for the big stuff they use floating heat sink modules with P-P wiring to rows of TO-3s. If it's a robust design it should never need replacement in its lifetime.
The really cheap stuff HAS to use sockets for selecting in test from bottom of the barrel batches of transistors /ICs.
I'm not criticizing the uA723 per se just your typical data sheet implementation of it. Besides rectification mentioned above you're completely missing finer aspects of the driver circuit and the tracking bias supply and how it changes the critical Vin /power requirements. You've tossed everything Astron has learned through the ages locked into a 723 and have replaced it with a beginners circuit. Those guys were actually pretty smart working with their hands /wallets tied into knots. added link for a power one hack >see TO-3 PCB images bottom of page http://www.djerickson.com/p1hack/
Last edited:
yup needs the best cooling b/c you'll have the worst efficiency of any modern fixed output DC supply found in the market place. what are your estimates of total wattage dissipated at 60A yet still providing full regulation at 80A?
Remember conducted heat rises so mount the TO-3 interfaces at the bottom of the heatsink panels. you'll probably need four fans externally mounted blowing air perpendicular to the fins. . Hotter running heatsinks are actually more efficient at removing heat . fullstop Your transformer will be cool but the rectifiers might be needing serious cooling. Since all that HS area wont be temperature tracking very well, you'll need much more emitter ballast than shown.
I think the best cooling is internally floating ( no TO-3 insulators ) heatsinks mounted in an air tunnel.
Last edited:
First, let me remind you kind Sir, that I'm posting my project design phase elements here to receive error analysis and recommendations for making it better. Some of your feedback has been appreciated and worth my time to read. But when you make silly comments that you can't back up, just to beat your chest, you lose credibility quickly. By design, Linear supplies have never and won't ever be more efficient than a switching design. Surely you're not comparing my system design / construction efficiency with the likes of Pyramid or Astron... because that would make you look like an idiot to say that theirs is more efficient than my project. And in case you aren't aware, the design I've presented here is not in production yet; there are many tweaks needed. For example, with a little voltage boost for V+, my efficiency dramatically improves. Mission accomplished, but thanks for your sarcasm.
The rectifiers are in the forced-air path within the chassis.
Yes exactly what I'm comparing this to , your power efficiency which is the driver for all the construction design decisions.
Surely you must have a calculation handy, before ordering the expensive transformer.
your claims of best in the biz, is the only chest beating going on here.
Last edited:
It's a 13.8v supply, so at 60A that's likely going to be close to 828 watts. Adding in the estimated power factor loss of approximately 1.4, that bring us to about 1160 watts. Even at high duty-cycle, and sized with a 1500VA transformer, that's not too shabby. Do you know of any 60A high duty-cycle supply in the market with those basic specs, that also have the cooling efficiency at sounds levels less than 40dBA that mine has? Please don't compare this desktop supply to a commercial rack-mounted supply blowing 4 80mm fans at 6500 at 80dBA. This sits on a workbench just like the Astron desktop supplies do (not in a rack isolated in a closet or data center).
How many times will you need to read, that I'm interested in boosting the V+ for added efficiency, before you stop getting hung-up on the early design revision schematics?
And, surely, you saw the smiley face next to the comment I made about creating the best linear that uses the LM723...yes? Let me help you; that meant I was smiling when I typed it. Stop being so condescending; what's the point?
how did you arrive at PF loss I never heard that term?
I usually calculate losses in the transformer , the rectifiers and finally the pass devices for a 1st level estimate.
at every point yer above what Astrons designs has got going on.
add a 1.4KW Variac get out jail card remains an option.
I usually calculate losses in the transformer , the rectifiers and finally the pass devices for a 1st level estimate.
at every point yer above what Astrons designs has got going on.
but the point remains, you've ordered a transformer so youre stuck, whether you raise the driver v+ or not.
add a 1.4KW Variac get out jail card remains an option.
Last edited:
the term, 'loss', was used to create a since of context in my response. I think your contribution has run its course. Thanks for your time.
Out of respect I'll respond to your question, but it will be the last time I do. As I've stated before, money is not a concern that I have, nor a factor in the design of this build. If I jumped the gun by order the 1500VA with a single winding at 17VAC, so be it. It's $220 for the Antek transformer. I spent more than that on the TO-3 mounting sockets. So, it begs the question, why the heck do you care; I thought we covered this already. Do you realize how simple it is for me to call the owner of Antek (he answers the phone often, oddly enough) and have him wrap another transformer for me? The very idea that you consider me, or anyone, 'stuck' with a transformer during the build of a prototype is crazy. Have a nice day.
Just curious, being a ham myself, what in the world are you going to run that requires 50amps at 13.8volts??
Pyramid makes a supply that seems to have good protection.
You can also look at the commercial supplies made by outfits like Lambda (back in the day) and see how they protected themselves.
I would advise against using any vented Lead-acid battery indoors. Too dangerous. Maybe sealed lead-acid type, but then unless you get them almost free are too expensive to be worthwhile. Bad idea to run a battery inside, can you say BOOOM??
_-_-bear
Pyramid makes a supply that seems to have good protection.
You can also look at the commercial supplies made by outfits like Lambda (back in the day) and see how they protected themselves.
I would advise against using any vented Lead-acid battery indoors. Too dangerous. Maybe sealed lead-acid type, but then unless you get them almost free are too expensive to be worthwhile. Bad idea to run a battery inside, can you say BOOOM??
_-_-bear
If low ambient noise is the goal, and a linear supply is the way you want to go (and not an SMPS) then perhaps a way to consider doing it, and assuming that size is not the issue, is to go "brute force".
That would be a ~1kva transformer core to start, run off 240vac, not 120vac. Now you have the requisite current available. Next you need to filter and control voltage.
Or do you? Yes to filtering. But IF your xfmr and line are stiff, then you could simply pick the iron to give you the desired voltage +/-10% after filtering. A very large cap bank could be eliminated with a suitably sized solid state "capacitor multiplier". The amount of vdrop in the cap multipliers is fairly low, so the size of the heatsinks will be modest. Fans not required at all. Overvoltage protection? Not required.
IF you decided to regulate a few volts - possible only IF you have ample current from the iron to back it up with - then the Vdrop across the regulators is low and the power in heat low, so the size of the heatsinks is not monster, and no fans required. Dropping a few volts might be 16volts down to 13.8. OVP could just be a comparator and relay to drop the B+ line...
The reason for these power supplies with lots of transistors to do regulation is that the manufacturers realized fast that to make a supply that does not weigh as much as a small safe, and to keep the cost down, that the way to go is to spec a medium current (VA) transformer, which is modest in size, at WAY too HIGH a voltage and regulate it down, burn heat not $$!
You don't need to do that, if you DIY and do not need portability.
If you DO need portability then it is more or less nuts not to use an SMPS and spend your time and money getting the spurious stuff filtered out, and end up with a <5lb supply - of course there may or may not be a fan involved with that...
My 2 cents - ca ching!
That would be a ~1kva transformer core to start, run off 240vac, not 120vac. Now you have the requisite current available. Next you need to filter and control voltage.
Or do you? Yes to filtering. But IF your xfmr and line are stiff, then you could simply pick the iron to give you the desired voltage +/-10% after filtering. A very large cap bank could be eliminated with a suitably sized solid state "capacitor multiplier". The amount of vdrop in the cap multipliers is fairly low, so the size of the heatsinks will be modest. Fans not required at all. Overvoltage protection? Not required.
IF you decided to regulate a few volts - possible only IF you have ample current from the iron to back it up with - then the Vdrop across the regulators is low and the power in heat low, so the size of the heatsinks is not monster, and no fans required. Dropping a few volts might be 16volts down to 13.8. OVP could just be a comparator and relay to drop the B+ line...
The reason for these power supplies with lots of transistors to do regulation is that the manufacturers realized fast that to make a supply that does not weigh as much as a small safe, and to keep the cost down, that the way to go is to spec a medium current (VA) transformer, which is modest in size, at WAY too HIGH a voltage and regulate it down, burn heat not $$!
You don't need to do that, if you DIY and do not need portability.
If you DO need portability then it is more or less nuts not to use an SMPS and spend your time and money getting the spurious stuff filtered out, and end up with a <5lb supply - of course there may or may not be a fan involved with that...
My 2 cents - ca ching!
Hi Bear,
Thanks for the comments. I agree, running banks of lead acid batteries indoors, even if vented, is not a risk I'm willing to take. 2M Amplifiers are what drove me to using large power supplies. After living with the poor quality and performance of Astron, Pyramid, and Tripp-Lite power supplies, I decided to build a better one, and have fun while doing it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.