THE 516F-2 PROJECT

P. Wokoun, KH6GRT (1/2004)

 



This is result of my attempt to produce an AC power supply equivalent to a Collins 516F-2 less expensively than I could buy one.

The need for a power supply became apparent by my working on Collins 32S3 transmitters and KWM2 transceivers. I would always have to make sure their power supplies came with them which meant hauling around lots of weight. I also recently acquired my own 32S3 and now needed a power supply for it.

Attemps to obtain one from e-bay at what I considered a reasonable price proved futile. Prices seemed to stay above $250 which seemed a little high considering the prices some transmitters sold for.

I had a replacement transformer for the Hallicrafters HT-32/37 series of transmitters. Both the Collins and Hallicrafters use a pair of 6146s so I reasoned the power transformers should be somewhat similar. I also had an LMB cabinet which looked like a poor man's Collins cabinet. So I decided to build a power supply around that Hallicrafters transformer in the LMB cabinet. I surely thought I could build one cheaper than what they were selling for.

Well, the end result is a supply, I think, that performs as well or better than a stock 516F-2. In the end it probably cost me just as much but, hey, I sure enjoyed building it and it does have a few more features than the 516F-2 has. What follows are some of the trials and tribulations I went through in arriving at the final design.

516F-2 PERFORMANCE


The target for my design was to be equivalent to a Collins 516F-2. The 516F-2 published specifications are:


High Voltage B++
Low Voltage B+
690 < B++ < 970
250 < B+ < 310
----------------
----------------
KWM2:
230 mA @ 800 VDC
210 mA @ 275 VDC
32S3:
240 mA @ 800 VDC
190 mA @ 275 VDC

 

Bias Voltage: -55 to -80 VDC

Filament Voltage: 6.3 VAC @ 11 Amps


516F-2 Front View516F-2 Bottom View

One thing I couldn't find in any of the tech manuals was how much current the 32S3 pulled in standby and the KWM2 pulled in receive. These values would have determined the minimum current pulled from the supply. Without these figures, the minimum current was considered zero.

I checked out a working 516F-2 supply with some fixed load resistors to see how closely it came to meeting its own specifications. The supply I used was almost stock, the only difference being the high voltage capacitors were replaced with 40 mfd units instead of 30 mfd units. This produces an equivalent high voltage capacitor of 13 mfd instead of 10 mfd. This is probably not a significant difference. If anything, it would probably improve its ripple reduction and regulation a bit.

The load resistors I used were 3448 ohms on the high voltage and 1430 ohms on the low voltage. They would load the high voltage to 232 mA at 800 volts and the low voltage to 192 mA at 275 volts. This current is just about what a nominal 32S3 or KWM2 would draw.

I powered the 516F-2 from both 115 and 120 volts AC and this is how the high- and low-voltage supplies performed:



115 VAC LINE
B++ HV (VDC)
B+ LV (VDC)
------------
------------
------------
NO LOAD:
1007
367
LOADED:
716
259

 

120 VAC LINE
B++ HV (VDC)
B+ LV (VDC)
------------
------------
------------
NO LOAD:
1060
385
LOADED:
746
273

 

One thing I noticed was how high the B++ high voltage went with no load. I would have expected its resonant choke input filter to hold it down to lessor levels. However, when loaded the supply did meet its published specifications.

BELLS AND WHISTLES

Some additional features I decided to include in my power supply:

1. A power-on relay, where the primary current to the power transformer is controlled by a set of high current contacts instead of the 32S3 or KWM2 power switch. The 32S3 or KWM2 power switch will only carry the current required by the relay coil which is a small fraction of an amp. This will lessen the load on the power switch contacts and extend their life.

2. Full metering of the high-voltage, low-voltage, and bias supplies. It's not really necessary but it's nice to watch how the supplies respond with loads. And, meters really impress visitors to the shack.

3. Resettable circuit breakers instead of fuses. I don't know if these thermal breakers will respond fast enough to protect anything except, possibly, in a short circuit condition. But I decided to give them a try and see how they worked out.

4. A test switch that operates the supply. It would be used instead of a wire jumper to turn it on and off. It just makes for easier operating during testing and troubleshooting.

5. A speaker to eliminate another box in the shack. A Motorola mobile speaker is used instead of a general purpose type.


Meter and Selector switchCircuit Breakers, Speaker Input, and Test SwitchSpeaker and Power/Soft-Start Relays

TUBES OR SILICON

The first concern I debated was whether to use tube or silicon rectifiers. I like the soft turn-on characteristic of tube rectifiers where the high voltage doesn't appear until after the filaments have had a chance to start heating up. This has to be easier on tubes than hitting them with high voltage and then start heating their filaments. Another consideration was the transformer itself. Transformers in the HT-32/37 rigs seem to have a higher than normal failure rate associated with a 5 volt rectifier filament winding shorting to another winding. Apparently the 5 volt windings were provided an inadequate margin of safety with regards to insulation. Thus I had a little hesitancy with using these 5 volt windings at high potential. I would also have to contend with the heat from the tubes in the enclosed cabinet. Ultimately, it came down to available space. When I began to physically layout the parts, it became obvious a couple of big rectifier tubes would take up too much chassis space. Thus I decided to go the silicon route.

After the silicon decision I decided to include a soft-start circuit to minimize turn-on stresses to the diodes. This soft-start circuit is where the initial line current is substantially limited until the filter capacitors have charged up. I found a delay of only about 1/4 to 1/2 second was more than adequate to provide this effect. After the delay full line voltage is applied to the power transformer. Another thought I had was to put another delay timer in the supply that would initiate the soft-start about 20 seconds after filament voltage was applied to the transmitters. I decided to drop this feature as 'excessive'.

Also after the silicon decision I included transient protection as much as possible for the diodes. This included a transient/surge absorber or varistor across the transformer primary, an RC snubber across the high voltage secondary, and RC equalizing networks across all the high- and low-voltage diodes. 3-amp diodes with a robust 200 amp surge rating were also used.


Power and Soft-Start RelaysB++ Diodes and Equalizing Networks100 ohm and 0.01/5KV HV AC Snubber

THE HIGH VOLTAGE SUPPLY

I originally intended to follow the Collins design and use a resonant filter choke filter in the high voltage circuit to limit the no-load voltage. I previously didn't have much success with this type of circuit, probably a result of the choke not being close to its expected value. But I decided to give it another try here.

The 516F-2 uses an 8 henry, 150 mA, 200 ohm choke in the high voltage circuit. It also uses a 0.05 mfd capacitor to resonate this choke to 120 Hz. If you calculate the inductance needed to resonate a 0.05 mfd capacitor at 120 Hz you come up with 35 henries. This is quite removed from the listed 8 henry choke. Could this be a swinging choke where the inductance is much higher at low currents? I had never seen mention of it being 'swinging' in any literature before. An 8 henry choke would need about 0.22 mfd to resonate it at 120 Hz.

Looking for chokes these days shows the availability is quite limited. Hammond Manufacturing seems to offer some reasonably priced units. Their closest to the Collins 8 henry unit is a 7 henry, 150 mA, 100 ohm unit, their model 159Q. The only problem is that its maximum operating voltage is only rated at 500 volts DC. To get a higher voltage unit I would have to get one of their enclosed units which has twice the volume and price. I decided to try the 159Q choke and look closer at its voltage limitation later.

Since I wanted to resonate this choke at 120 Hz while drawing bleeder resistor current, I needed to know its inductance at about 20 mA. Communicating with Hammond I found they couldn't provide this data, just its inductance at rated current. Now, measuring inductance with various DC currents is not something your common bridge or meter will do. Searching the internet I came across an interesting article that appeared in EDN Design Ideas on July 6, 1995 by H. B. Farensbach. The article showed how one could measure inductance with DC superimposed without it being too complicated. I played around with that circuit and was able to develop some inductance vs. current curves for the Hammond 159Q choke.

(That EDN article may no longer be accessable at EDN archives; click here for the Farensbach article in .pdf file. Farensback talks about making a special transformer; I just used a 5 VAC, 3A filament transformer and a test frequency of 120 Hz which worked just fine. I used an oscillator driving a PA amplifier with a 70 volt output to drive the filament transformer, however, I'm not sure I really needed the amplifier but didn't pursue that further.)

While developing the inductance curves for the 159Q choke I also decided to see how varying its air gap changed its inductance. The choke comes with a nominal 0.02 inch gap and I took additional measurements with gaps of 0.01, 0.003, and zero inches. I measured the stock choke inductance at about 6 henries rather than its rated 7 henries. However, it did stay fairly constant up to about 250 mA. Naturally, decreasing the air gap showed increasing values of inductance at low currents and subsequently less inductance at the higher currents. With a decreasing air gap the core is saturating with less current. An air gap decreases the inductance while increasing the operating current before the core saturates.

One interesting effect was with the zero air gap. Here the 159Q choke has an inductance about 15 henries up to about 25 mA decreasing to 2.5 henries at 150 mA and just under 2 henries at 200 mA. This high inductance, while not as high as the critical inductance with only bleeder resistor current, should limit the unloaded voltage quite well. As the load current increases and lowers the inductance, it would operate even more below the critical inductance and start approaching a capacitor input filter. I decided to test the choke in this mode as a 'swinging' choke before trying to resonate it.

The high voltage was tested using an unmodified choke as well as the swinging choke. The results were:


 
7 Henry Choke
15-2.5 Swinging Choke
Stock 516F-2
 
------------
--------------------
------------
NO LOAD:
1004 VDC
917 VDC
1060 VDC
LOADED:
730 VDC
734 VDC
752 VDC
VOLTAGE DROP:
274 VDC
183 VDC
308 VDC

 

Out of curiosity I shorted the swinging choke under load to see what a pure capacitor-input filter would provide. The voltage went up from 734 to 1007 volts. A swinging choke with even less minimum inductance would keep the loaded voltage even higher here. However, with these gratifying results I decided to forget any attempts at a resonant choke filter and use it as a 'swinging' choke-input filter.

That 500 volt maximum rating on the choke now gave me a little concern. In a transformer, exceeding the voltage rating can result in a winding-to-winding breakdown or a winding-to-core breakdown. Since a choke is a single winding at essentially the same potential, the maximum voltage rating would only involve a winding-to-core breakdown. Normally the core is grounded for safety's sake. But if the core is left ungrounded, any breakdown must occur between the winding-to-core AND the core-to-ground. The breakdown voltage is increased because of the core-to-ground isolation. Since the choke was going to be installed within an enclosed cabinet, I elected to isolate its core from ground with fiberglass board and appropriate warning labels. I did include a circuit with a neon bulb that would have a warning glow should the choke develop any winding-to-core leakage or short.

I also increased the filter capacitance from that used in the 516F-2 to compensate for the decreased choke inductance at the higher currents. I used capacitors that give a total high-voltage capacitance of 33 mfd whereas the 516F-2 uses 10 mfd.


15-2.5 H Swinging Choke and Neon Warning IndicatorHigh Voltage Capacitor Assembly (9 X 33 MFD/400 VDC)


LOW VOLTAGE SUPPLY


For the low voltage supply the 516F-2 transformer has a much higher output voltage than the transformer I was using: 425 volts vs 270 volts. I followed the 516F-2 filtering fairly close, using a 7 henry input choke and a 1.5 henry output choke whereas the 516F-2 uses a 8 henry input choke and a 1 henry output choke. With the lower transformer voltage to the filter, it was obvious the DC output voltage was going to be too low. I had to use a capacitor input filter to compensate for it. I settled on 8 mfd for the input capacitor that produced almost identical results with the 516F-2 between unloaded and loaded conditions as follows:

 
LOW VOLTAGE, NO CAP
LOW VOLTAGE, 8 MFD
STOCK 516F-2
 
-----------------
-----------------
------------
NO LOAD:
292 VDC
383 VDC
385 VDC
LOADED:
225 VDC
278 VDC
273 VDC

 

I also increased the output capacitance to improve the filtering.

Low Voltage Diodes and CapsLow Voltage 7 H ChokeLow Voltage 1.5 H Choke


BIAS SUPPLY

The only problem I have ever had with a 516F-2 bias supply, other than the selenium rectifier originally supplied, was its variance with line voltage. This directly caused the idling plate current to vary with changing line voltage. Granted this is not a real big problem! I generally used the 516F-2 bias circuit except for the addition of a regulating zener diode and an additional filter capacitor. It's output now remains constant with varying line voltages because of the zener and the additional capacitor has reduced ripple voltage on the bias down to very low levels.

Bias Circuit ComponentsFront Panel Bias Adjust Pot




MISCELLANEOUS TOPICS


Another item added to the supply is a small 12-volt fan about 2-inches in diameter that exhausts heat out of the cabinet from around the 25 watt bleeder resistors. This fan and the power-on lamp are powered from the rectified 5-volt windings. No filtering is used after the rectifiers so they only provide about 8-9 volts equivalent DC. This voltage lets the fan run at less than full speed with a lower noise level and increases the lamps's life by using slightly less than rated voltage.

Fan from Rear PanelFan and 25 Watt Bleeder Resistors


I also elected to use a separate filament transformer rather than use the power transformer's filament winding. This was to eliminate a load on the power transformer that would let it run cooler, hopefully lengthening its life and maybe improving its regulation a little.

6.3 VAC 10 Amp Filament Transfomer




The meter originally started life as a nice-looking Radio Shack Vu meter I had around. These Vu meters are nothing more than a sensitive DC microammeter with a diode bridge and a couple calibrating resistors tossed in. After measuring the movement's internal resistance and the current needed to produce full scale deflection, the dropping resistors needed for the different voltage scales were easily calculated. I used a 1000 volt scale for the high-voltage, a 500 volt scale for the low-voltage, and a 100 volt scale for the bias supply. A little artwork on the Visio program and I quickly had a new scale glued onto the backside of its faceplate. Multiturn trimmers were included to calibrate the meter right-on.

Ex Vu Meter With New ScaleMeter Scale Adjustment Pots


Teflon insulated wire was used throughout because I like working with it and its insulating capability is outstanding. It's nice to solder it and not have the insulation melt back.

All the electrolytic capacitors used have a 105 degree centigrade rating rather than the common 85 degree rating. I think a few cents more for the higher temperature ones will pay off with less problems in the future.


SUMMARY

In summary, the following table shows how well my completed supply compares to the 516F-2:


	
 
MY POWER SUPPLY
516F-2
 
-----------------------------
--------------------------------
 
VOLTS DC
VOLTS RIPPLE P-P
VOLTS DC
VOLTS RIPPLE P-P
 
---------
------------
---------
------------------
B++, NO LOAD:
884
2
1089
6
B++, LOADED:
755
20
779
20
DROP = 129 VOLTS
DROP = 310 VOLTS
       
B+, NO LOAD:
373
0.02
388
0.2
B+, LOADED:
283
0.05
286
0.4
DROP = 90 VOLTS
DROP = 102 VOLTS
       
BIAS:
46-88
0.02
55-74
0.4

These readings were taken with a constant 117 volt AC line voltage with both the high- and low-voltage supplies loaded.

So, did I succeed in keeping my cost down by building it? No. The cost for new parts actually included in the power supply ran about $225. This doesn't include anything from my well-stocked junk box which included the power transformer, meter, cabinet, hardware, terminal strips, speaker, wire and whatever else I picked up in years past. Granted some was for features I 'added in'. But from a long time builder, I'd still do it again!


Front View of Completed Power SupplyRear View of Completed Power SupplyInternal Top ViewUnder Chassis View 1Under Chassis View 2

PS516 Power Supply Schematic in PDFPS516 Power Supply SchematicPS516 Power Supply Parts Listing in PDF


Eventually I'll be putting some labels on the panels.



TOP BACK HOME

END