The AM Forum

I'm looking for ideas or a schematic for a linear voltage regulator.  Input will be 160VDC from rectified mains and output needs to hold steady at 134 Volts at up to 5 amps with a fluctuating load.  Any ideas??
73s  Nigel

Define "steady". Do you mean 134.00 volts +/- nothing, or would say plus or minus 3 volts do as well?   

Nigel,

   Perhaps you can add some further information. Direct rectification off the mains (120v)  if full wave means the load must float. Half wave could be ground referenced, but at 5 amps, that would be a mess with sag between pulses. Is an isolation transformer possible?
   Then there is a need to be short circuit protected, or not. Does the regulator need to survive an overload? A simple fet source follower offers some degree of regulation and ripple reduction. How pure a DC do you need, and what are the regulation requirements? Would a 5 volt drop from 0-5amps suffice?

Jim
Wd5jko

What I have is 120V 60 hz wall plug power feeding a full bridge rectifier to a 5000 mfd capacitor.  The transformer is outside up the pole feeding the house.  It provides about 160V DV with no load.  At 4 amps there is a lot of 120 Hz ripple and since the load is a modulated RF deck the load is not constant causing further voltage fluctuations.  I want to regulate this to some lower voltage below the ripple level, say 130 Volts but 125 would do its not that critical, just whatever it takes to get steady DC.  To put this in context my station 13.8 Volt supply uses a LM723 regulator chip driving 3 pass transistors.  It is rock steady at 13.8V when transmitting.  But I don't know how to scale this up to 135V.
Nigel

   If I understand what your doing (IF), then you will have two grounds (AC POWER) and the transmitter power return. If these are in common with each other, you cannot use a full wave bridge since the (-) output of the bridge is at a different potential that the AC power Neutral wire. By CODE, neutral wires connect to to ground at the power entry panel.

  It would be better to get an isolation transformer. Heck Control transformers such as this are quite inexpensive:

http://www.ebay.com/itm/Hammond-127414-x13550057-02-Control-Transformer-500VA-Pri-200-230-460V-Sec-115V-/271585161595?pt=LH_DefaultDomain_0&hash=item3f3bbcb17b

  Then if your RF amplifier pulls lots of current in a fault state, what is the means of current limit?

Jim
Wd5JKO

If you have 240V, then you can do full wave center tapped. For safety the power system must float and be isolated from the chassis except at RF.

If you want to float the power system and use a bridge and capacitor or inductively couple the RF, OK too. You have to isolate the chassis from B- for everything except RF. TV sets did that floating chassis thing a lot, and 115-135VDC was very common regulated B+ for a TV.

I agree though that you should use transformer isolation, because a small error in design can build in a potential hazard that could appear with no warning and kill someone.

Usually the solution in a TV set was a large dropping resistor across a series regulator, so as to use a smaller regulator and let the resistor take the bulk of the load. Easy to scale that to an all-pass-transistor scheme. Check some Panasonic color TV diagrams.

Other sets used an SCR to regulate the voltage, see RCA designs. It is simple to use two SCRs in place of two of the diodes in a bridge to do the gross regulation and then an active filter and subregulator to do the fine regulation.

The PP-4763/GRC POWER SUPPLY did this SCR + active filter thing to make 28VDC at 50A with +/-0.5% regulation.

These ideas are simple to scale to your voltage or current. I'll what i have for diagrams and put them up. Load up your present diagram, I am sure it is simple to go from there.

The one with the TIP50 has a nice explanation page: http://www.tubecad.com/2006/11/blog0087.htm

The action of the two in series there is a little odd. It's not so good at really regulating.

Messed around in LTSpice. Components were very limited, i.e. the triple Darlington of 2N3055's.

But this different scheme of the above works in there, and regulation of 134V is shown at 0.5A and 5A, good. So, with better choice of the pass transistors and the driver, it should do as you want.

Opcom I thank you for that.  Excellent article which brought things to light.  I especially like the opamp / FET design below the one you modeled. I just couldnt get my head around it but now I understand.  Will cobble one up today and test with a variac input and resistive load.

BTW this is the last step in my AM TX project which is all strung together on the bench pumping 100W (400 PEP) into a cantenna fed with Shania Twain audio from an MP3 player.  It sounds fantastic on my SDR RX except for the PS ripple which I have not been able to completely null out. So will try some voltage regulation and take a small hit in efficiency. Im striving for Borg perfection with simplicity. Resistance is not futile its E/I :-)
73s  Nigel

The easiest way to skin this cat is to use a "cap multiplier" via a high voltage mosfet to act as the pass device.

I'm unclear as to if you are aiming at a lightweight compact PS or not. IF not, then a set of cap multipliers, one before and one after a choke will do the job very nicely. Maybe better than a typical vreg since it will handle bad situations somewhat better...

Since you don't really need precision voltage regulation, this is a good way to drop the hum and ripple with a minimum parts count.

Do not neglect to use the protection diodes around any regulator circuit that is set up for HV. It's easier to get away without these diodes at low voltages.

                           _-_-

Back in 2003 I made several SS regulators for an old Sherwood Tube Amplifier. They were all high voltage and low current. The schematic attached could be adapted for use here. I'd pick a big Iso-Top 500V FET for the series pass. Then in the circuit eliminate D1, and replace R1 with 2-3 ohms. The idea is to keep the FET in the hard on region when the regulator is shorted out. The idea is an upstream fuse will blow before the FET. The zener D5 gate to source on the FET could come up to 12 or 15v too. The use of a CCS feeding the zener string really adds to the ripple reduction, and improves the line voltage regulation.

Here is what I originally wrote about this regulator:

"My first regulator was for the FM section of my Sherwood S8000 FM and 30 watt / channel tube amp. The stock voltage had significant ripple, and poor regulation. Some of the ripple came through with the audio, and the poor regulation sometimes disturbed the "lock" in the FM Multiplex adapter circuit. The regulator here provides over 60 DB of ripple reduction, and offers fair open loop regulation with a source follower fet series pass with very clean dc on the gate. The 4 ma CCS on the input helps maintain constant zener current with varying input voltage. This aids the ripple reduction, stabilizes the output voltage some, and gives the circuit decent line regulation. The zeners do warm up a bit, and the output drifts upward about 5 volts during the first 5 minutes of operation."

Like the CC powered reference.   Nice thing about Jim's approach is, more soup simply means more FETs, more heat sink and more air. 

Based on the good inputs here I bread boarded a few regulator circuits using parts I had on hand.  They all worked well. But the more I got into it the simpler it became. I ended up with 3 40V zener diodes in series fed with a 2.2k resistor from the +160 volt rail. The diodes are bypassed with a 100mfd cap giving 120volts clean DC.  This feeds the gate of a IRFP260 FET with the drain at +160 the source to the load is about 117VDC .  The source required .56mfd bypass to stop mild oscillation. This puts the output well below the PS ripple level at 4 amps load current. I will get some different zeners and bring the O/P up a bit but for now its just great.  Load regulation is .3V going from zero to 4 amps better than I expected. The FET gets warm but a 2.5 inch cubed CPU heat sink with the fan at half speed easily sinks the heat.  Though its loosing some efficiency in heat its still way better than an analog modulator. Hard to believe now it could be that simple.
73s  Nigel