Bad or Good Idea; 13.8V Linear DC PS for HAM Radios using 1 of these regulators?

[Nipkinz]

Hi gang,

I'm considering putting together a design schematic for a linear DC 13.8V power supply used for Ham Radios. Many linear DC supplies for Ham Radios make use of the famous LM/uA-723 voltage regulator chip. After reading through a few posts here related to the 723, I'm considering using a different regulator. For my Ham Radio requirements, typical load current will range between various radios between 9A-23A, therefore I'm planning for a PS capable of providing up to 40A 100% duty-cycle max. However, typical max current load will likely never exceed 25A at 100% duty cycle.

Strongly desired regulator specs would include:
- Adjustable output voltage (either via built-in pin(s) or external resistors)
- LDO
- Low output voltage noise (perhaps <80µVrms)
- High ripple rejection (perhaps >50dB (100 Hz)
- Input voltage range of 18v-35v (wide range because I've never worked with a LDO regulator and not certain how much drop is associated and/or specified for stable and safe operation)
- Current limit protection (short and reverse protection)
- Over-voltage protection adjustment to protect the radio(s).

This is a big ask, and is my first forum post, so I'll apologize now for saying or asking stupid things, *grin*.

So far, in my search, I've come across two possible regulator candidates. What are your thoughts and recommendations?

1. TPS7A4701-EP:  https://www.ti.com/lit/ds/symlink/tps7a4701-ep.pdf?ts=1724489932033
2. UC2834M:  https://www.ti.com/lit/ds/symlink/uc2834m.pdf?ts=1724513506097&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FUC2834M

I appreciate any input you provide, have a great day!

[Wallace Gasiewicz]

Building a 30 amp Linear PS is not hard but probably quite expensive. I would look at the Astron or Pyramid PS and find a schematic.   
But the most expensive part will be your Transformer.     
If you just need a PS for regular use I would look at Hamfests. Even a broken one would be probably easily fixed, as long as the transformer is OK.     You will probably find that an entire used PS will be cheaper than buying the transformer alone.
I use a Yaesu FP 1030A PS for testing radios. Since I test broken radios I need something with good safety features and an ammeter. I bought it broken at a Hamfest for $20 and fixed it in about an hour.     

https://www.arrl.org/hamfests-and-conventions-calendar

[Nipkinz]

Building a 30 amp Linear PS is not hard but probably quite expensive. I would look at the Astron or Pyramid PS and find a schematic.   
But the most expensive part will be your Transformer.     
If you just need a PS for regular use I would look at Hamfests. Even a broken one would be probably easily fixed, as long as the transformer is OK.     You will probably find that an entire used PS will be cheaper than buying the transformer alone.
I use a Yaesu FP 1030A PS for testing radios. Since I test broken radios I need something with good safety features and an ammeter. I bought it broken at a Hamfest for $20 and fixed it in about an hour.     

Hi Wallace, thanks for responding, I really appreciate that. To clarify, I'm not seeking advice on buying a power supply, I asked which regulator would be well-suited for building my own power supply and included two potential candidates for consideration. Thanks again for responding.

Cheers 

[bingo600]

Hi Wallace, thanks for responding, I really appreciate that. To clarify, I'm not seeking advice on buying a power supply, I asked which regulator would be well-suited for building my own power supply and included two potential candidates for consideration. Thanks again for responding.

Cheers

What an attitude for a 2'poster 

[Nipkinz]

Hi Wallace, thanks for responding, I really appreciate that. To clarify, I'm not seeking advice on buying a power supply, I asked which regulator would be well-suited for building my own power supply and included two potential candidates for consideration. Thanks again for responding.

Cheers

What an attitude for a 2'poster 

What? I replied with a clarification to the responder who missed the context of my post. I think you're seeking drama where none exists. Interestingly, now there are two irrelevant responses to my post, thanks for your time.

[ArdWar]

I'm not sure with your design requirement. 35 V input, 13.8 V output, 40 A capacity. That's potentially 850 W heat to dissipate.
That's... a lot

Forget TPS7A4701 (unless you want to parallel 300 if them), you need external pass element. Even then you may still need to parallel lots of them.

[coromonadalix]

with all your needed specs ans so on, better buy an already made one,  pfc correction ovp ocp thermal etc ...

i dont think even with answering your 2 choices,   you'd catch  on how it's difficult to get theses specs, and reliably supply your ham radio and protect it against failures

Ham radios have a tendency to kick more current (s)  for a brief moment when you emit .. and some  smps  may hate this ..

and some rigs could catch the switching frequencies of the DIY smps you want to build ...

i have a friend who has rig Ham stuff and we had to test many smps  before getting to real reliable brands, and have peace,  low to zero harmonics etc ...

[madires]

For 40A you'll need several pass transistors in parallel (each one with an emitter resistor) and a driver transistor (or two). It's not going to be an LDO. Also, as already explained by ArdWar, the heat dissipation for an input voltage up to 35V will be insane.

[Nipkinz]

I'm not sure with your design requirement. 35 V input, 13.8 V output, 40 A capacity. That's potentially 850 W heat to dissipate.
That's... a lot

Forget TPS7A4701 (unless you want to parallel 300 if them), you need external pass element. Even then you may still need to parallel lots of them.

Hi ArdWar, thanks for responding mate. I don't have a design requirement of 35 V input to the regulator, I noted a range of regulator input voltage specs so that I can then choose from a broad list of potential devices. For example, some regulators max input V caps out at 20V, which, likely wouldn't be quite enough; perhaps 2'ish V short for example unless a LDO regulator could operate between 16V-18V for example, but I've never designed with LDO, so it's uncertain at this time. I can't imagine pushing 35V into a regulator with only needing 13.8V for the ham radios,  and wouldn't design a PS with that high of an input. Would you mind helping me understand why I would need 300 TPS7A4701's? When I read through the data sheet, a single TPS7A4701 will push enough current to drive pass transistors just fine. Perhaps I'm not understanding something. 

[Nipkinz]

with all your needed specs ans so on, better buy an already made one,  pfc correction ovp ocp thermal etc ...

i dont think even with answering your 2 choices,   you'd catch  on how it's difficult to get theses specs, and reliably supply your ham radio and protect it against failures

Ham radios have a tendency to kick more current (s)  for a brief moment when you emit .. and some  smps  may hate this ..

and some rigs could catch the switching frequencies of the DIY smps you want to build ...

i have a friend who has rig Ham stuff and we had to test many smps  before getting to real reliable brands, and have peace,  low to zero harmonics etc ...

Hi coromonadalix, thanks for responding mate. My list of specs are noted as desired (not absolutely required). I've designed with the uA723 previously, but I'd like other regulator options and leave the uA723's behind me. The uA723 is a great regulator but I'm hoping there are modern / better options these days perhaps? You mentioned SMPS a few times in  your response, but as stated I'm designing for linear, not SMPS. You're totally right about the peace of mind with a DIY Power Supply for the expensive Ham Radios out there today. I just saw that ICOM is releasing a brand new IC-7760 for nearly $6,000. Designing for ovp crowbar is the easy part, the LDO with enough input V spec and low noise might be the more challenging aspect of the project.

To clarify, I would like to use the least amount of regulator input voltage required for stable operation of this linear 40A 13.8 supply. (keep in mind that nominal operation of this supply would be 25A full duty cycle worst case scenario). And you're right, cooling and proper PCB design are both paramount to success. Let me know if you have any suggestions for regulators you feel might come close to the specs noted above. Have a great day, cheers! 

[iMo]

..
To clarify, I would like to use the least amount of regulator input voltage required for stable operation of this linear 40A 13.8 supply. (keep in mind that nominal operation of this supply would be 25A full duty cycle worst case scenario)..

The 40A 13.8V is a good design goal. What will be your biggest challenge is to satisfy the "least amount of input voltage.." requirement..
Your RX current could be 2-3A easily, and your TX current up to 25A with 100W (it depends on the operating mode). Your transformer's output voltage will jump around based on those loads and the construction of the trafo.

The minimal voltage could be, say, 13.8V+5V+2V+3V = 23.8V (output, pass transistors, diode bridge, cap filter ripple) at full load, it will jump up to perhaps 30V or more at the low load based on the transformer construction (the core material, geometry, etc).

So the power loss at the electronics itself (30V-13.8V)*2A = min 32W at RX, and (23.8V-13.8V)*25A = aprox 230W at TX (or aprox 380W at 40A). Not counting the loss at the transformer.
That would be a good result, imho.

[Nipkinz]

The 40A 13.8V is a good design goal. What will be your biggest challenge is to satisfy the "least amount of input voltage.." requirement..
Your RX current could be 2-3A easily, and your TX current up to 25A with 100W (it depends on the operating mode). Your transformer's output voltage will jump around based on those loads and the construction of the trafo.

The minimal voltage could be, say, 13.8V+5V+2V+1V = 21.8V (output, pass transistors, diode bridge, cap ripple) at full load, it will jump up to perhaps 30V or more at the low load based on the transformer construction (the core material, geometry, etc).

So the power loss at the electronics itself (30V-13.8V)*2A = aprox 32W at RX, and (21.8V-13.8V)*25A = aprox 200W at TX (or 320W at 40A). Not counting the loss at the transformer.
That would be a good result, imho.

Hi iMo, thanks you for helping me keep my sanity; I was struggling with loss-assumptions because I haven't even put this design into schematic format yet. I was waiting to do that if/when I found a great linear regulator that I could get passionate about. A mistake made (there was more than one, *grin*) on my previous uA723 regulator-based linear DC supply was that I went too far and unnecessarily overboard on the bridge rectifier and filter caps. That PS had too much V drop/loss before getting to the regulator, and also experienced slow-start design deficiencies.
The UC2834M is starting to look like it's currently top of the short-list. Thanks again mate, cheers!     

[iMo]

I updated the calculation above - the ripple at the capacitor (after the diode bridge) would be much higher than 1V I had there previously, so I put there 3V, what is still a pretty optimistic value, imho.
The design of such a power supply is not about some mysterious chips but rather the entire concept. In our ham club we have there a perhaps 40-50y old linear psu (all others are switchers), people from time to time use it with modern 100W output all mode all band rigs, large heavy box with perhaps 6 pass transistors mounted on a large heatsink.
The voltage regulator itself (not counting the diode bridge, pass transistors and crowbar with a thyristor) could be almost anything (will work with say 4-5 transistors and a zener, as well as with any modern chip), what is quite tricky to design is all that stuff around it - the heavy metal.

[Nipkinz]

I updated the calculation above - the ripple at the capacitor (after the diode bridge) would be much higher than 1V I had there, so I put there 3V, what is still a pretty optimistic value.

Aye, roger that. My last PS needed 24.2V feeding the uA723 at full load, and jumped to nearly 36V no-load. That was an expensive Transformer and the PS was built for much higher power than the 40A design I'm considering building now. Ripple will be an issue that I still need to solve. 

[NiHaoMike]

How about a switching regulator with a small LDO in parallel so that the switcher doesn't actually switch at low current?

[Nipkinz]

How about a switching regulator with a small LDO in parallel so that the switcher doesn't actually switch at low current?

Hi NiHaoMike, interesting and I have never even considered that idea. I'll check into this; very cool, thank you!

[floobydust]

The pass-transistor array is typically a triple-emitter follower with associated high voltage drop, say 3V. This is overcome with the extra DC bus voltage, but also makes more heat.
Instead I would consider using an aux transformer/winding to power the regulator circuit.

Had to laugh, just repaired a 20A 13.8V PSU that uses TL431 for voltage regulation.
I think it's not about the regulator IC - a vanilla voltage reference, voltage-error amp, current-sense op-amp. Nothing that discrete IC's can't easily accomplish and are far easier to source.
I see car batteries used for transient high currents as a way to lower the PSU requirements and cost, as well.

[Nipkinz]

The pass-transistor array is typically a triple-emitter follower with associated high voltage drop, say 3V. This is overcome with the extra DC bus voltage, but also makes more heat.
Instead I would consider using an aux transformer/winding to power the regulator circuit.

Hi floobydust, thanks for this recommendation. I really like the idea of an aux transformer for the regulator. Years ago I consulted with a PS design guru who suggested the same thing. I kick myself often for not taking that advice. 

[Wallace Gasiewicz]

Flooby beat me to it.....     
We use old Military linear PS at the Radio Shop. They are so big that one person cannot lift one. More than 100+ Amps.   
Lots of Pass Transistors. So I am familiar with these things. I have also repaired lots of Astrons etc.....
Here is what I think:  The first step is deciding what sort of Transformer you are going to use.       
The idea is not to burn off power unnecessarily.  The Headroom should not be huge.     
Power for the regulatory circuit in current Linear Supplies these days is supplied by another secondary on the transformer. Essentially another transformer.  Lots of the Lab Grade PS use this type of circuitry.     
Here is an example of this type of circuit:    file:///tmp/mozilla_mint0/yaesu_fp-1030a_sm.pdf
The older Astron PS used a slightly different approach using the 723:     https://www.ameradio.com/doc/Astron%20RS-35A%2C%20RS-35M%20schematic.pdf   

Notice the voltage for the regulator is not coming from the same transformer taps as the main power output.

But the idea of regulating the main power with big pass transistors remains the same.  You really do not want to down regulate 35 volts to 13.8 Volts, unless you are cold and need a space heater. The voltage in the "housekeeping" supply is not affected by the main power output, but separate and does not change with load on the output.

[Nipkinz]

Power for the regulatory circuit in current Linear Supplies these days is supplied by another secondary on the transformer. Essentially another transformer.  Lots of the Lab Grade PS use this type of circuitry.     
The older Astron PS used a slightly different approach using the 723:     https://www.ameradio.com/doc/Astron%20RS-35A%2C%20RS-35M%20schematic.pdf   

Notice the voltage for the regulator is not coming from the same transformer taps as the main power output.

Hi Wallace, with your experience repairing Astrons, you might have seen the following schematic, but perhaps not yet. Astron has made a revision to their big boys in 2020. Let me know what you think, although my schematic will differ of course because I'll be using a different regulator, but concepts remain with regard to pass transistors, etc.  Latest Astron VS70M Schematic PDF: https://drive.proton.me/urls/NPX1TDMHJ8#kVxw7qDIko3T

Check out what they're doing with the 7.2 rails; I like it.

[iMo]

Also Harrison type of linear PSU architecture (3 floating secondaries) may fit your goal..
Like this one - couple of transistors, btw..

[schmitt trigger]

A salvaged AC motor, a salvaged car alternator, belt and pulleys, a car battery, and voila!

Hope people realize this is a tongue in cheek answer.

[Wallace Gasiewicz]

Astron has always put out a good product.  Their new schematic is just fine. Similar to older units in the transformer topology.  I kind of like the Yaesu design as it lends itself to different transformers rather than one just made for Astron.  All you need is a big transformer for the 13.8    volt output and any little transformer will do for the "housekeeping"     
With this topology, you can probably more easily source transformers.

As to the drop in voltage needed in the main power output, as I recall the huge 100 plus Amp PS we use start with only about a 16.5 volt after rectification that is brought down to 13.8 volt output. Not Much waste. But very large transformers.   

Also, just a thought, since MosFETS are so cheap and can handle lots of power, why not use mosFETS instead of bipolar pass TO3 transistors. Easier to mount on a heat sink also.  Probably require much less drive so probably a whole stage could be eliminated.  Good TO3 power transistors are not cheap.

[ArdWar]

Would you mind helping me understand why I would need 300 TPS7A4701's? When I read through the data sheet, a single TPS7A4701 will push enough current to drive pass transistors just fine.

I guess you can use it that way. Although that will be a bit weird choice since once you use discrete pass element all of their performance/protection bell and whistles are thrown out of the window.

[T3sl4co1l]

For the full voltage range and current as stated, you need to dissipate up to 848W.

That's a big heatsink, and at least several large transistors, probably a dozen modest sized ones.

The specs are absolutely achievable by SMPS alone -- no need for pre/post reg.

For simplicity, I might opt for a hybrid design, iron transformer + rect into buck regulator; LM25119 and similar parts would be a reasonable choice.  Power factor is a concern, especially at these power levels (13.8V 40A is 552W; at 90% overall efficiency, 0.5 PF and 120V input, that's 10.2A RMS already!), and I mean just in practical terms.  Using a proper offline SMPS is highly attractive.

As it happens, I have a stock design (mains to  24V 30A, LLC resonant) I could apply to such a purpose, or buy a unit ready made.  It tends to be a bit noisy in the audible spectrum (would not meet the ripple requirement, unless maybe the requirement is low frequency only) so the buck stage would still be necessary, rather than designing it for 13.8V exactly in the first place.

I do wonder just how sensitive the load is, if it really needs to be 13.8(00..) V and low ripple, or if anything from 12 to 15V would do, and what the effect actually is.  I.e. are you asking for precision out of an abundance of caution, or is there a real and present need for it?  Compare class A/B to C/D (maybe E) amplifiers for example; SSB vs. FM/PM use, perhaps.  Linear amps generally should have better PSRR, give or take biasing anyway, and it may be less effort to fix the latter than to regulate and filter the former.

Even with an SMPS + postreg architecture, the required voltage drop is much smaller than for raw rectifier output.  Even if you have a tap changer or variac to compensate for mains fluctuation, the peak-to-peak ripple is still enough to need a hundred watts or so of dissipation with the rectifier.  Very low overhead can be used with an SMPS, but mind that loop gain will be lower, and compensation more difficult, as the transistor's operated near or in [voltage] saturation.

Tim