The constant voltage machine

[Hugo Holden]

I have just put a detailed article on a "Constant Voltage Machine" (called a "machine" because it has a motor, gearbox and moving parts) which I have just finished making. I built it to power my SOL-20 computer. But it would work for any S-100 computer or any other 115V operated appliance of < 200W rating that would benefit from a stable controlled line voltage. It has an option for including the computer's power supply in the feedback loop.

The logic behind building the machine in this manner is explained in the article, along with the construction details and schematics, it is not possible to explain it all on one post, the article is here:

http://worldphaco.com/uploads/THE_CONSTANT_VOLTAGE_MACHINE.pdf

 

[Chuck(G)]

Are ferroresonant (CV) transformers still produced?

 

[Hugo Holden]

Are ferroresonant (CV) transformers still produced?

Yes I'm pretty sure they are, but I have never liked them much. One I tried in the past had what I would call "loose" voltage regulation and lacked the ability to set the output level to some specific voltage.

For the application of minimizing heat dissipation in the computer, it is useful to be able to set the output voltage to a specific value and also have it stable after that as well.

 

[Chuck(G)]

Isn't that what a well-designed modern power supply does? My bench supplies certainly do that. In fact, if you want really good regulation, you use 4 wire configuration (2 supply+2 sense).

 

[Hugo Holden]

Isn't that what a well-designed modern power supply does? My bench supplies certainly do that. In fact, if you want really good regulation, you use 4 wire configuration (2 supply+2 sense).

Yes that is right, but the SOL-20 computer requires an AC input to use its existing regulators and the S-100 cards require their standard voltage inputs, so short of stripping the entire computer power supply away and replacing it with a modern SMPS version with sense wires, the only way to achieve better stability and lower heat dissipation is to "pre regulate the regulators", which is the aim of The Constant Voltage Machine.

What it means is that using this machine, you can run the pre-regulator voltages in the SOL at the lowest possible level, to reduce heat dissipation, without dips or troughs in the incoming line voltage resulting in ripple breakthrough on the regulator's outputs, being both the typical 7805 regulator's on the S-100 cards and the Sol's own 5V regulator too.

If you did this with a Variac on its own, there would be times where the line voltage dropped low enough to get ripple breakthrough in the regulator outputs. As noted in the article I have been trying this with 7.6V on the sol backplane (just enough to satisfy the 7805 regulator conditions), but without issue, because when the line voltage fluctuates low, the machine corrects for it. This results in the lowest possible heat generation in the computer.

 

[Gary C]

I know there are digital methods of providing a pure, controlled sine wave output

But i love that, beautifully made.

The coupling looks familiar. Seems the same as ones used by Bristol Babcock for our hand/auto motorised potentiometers at work.

Love the use of the 741 (well, equivalents), I built a three term PID controller out of 741's in the 80's for a college project. Wish I knew where it was.

 

[Chuck(G)]

My experience with early S100 cards told me that even with linear regulators and the standard 8V line supply, manufacturers were awfully cavalier about loading those poor regulators. MITS was notorious for this, even adding shunt resistors across the regulators to keep them from burning up. I've never played with a SOL, but I suspect, given the age, that they were little better.

 

[lowen]

Hugo, nicely done. Motorized variacs are fairly standard high power voltage regulators in industrial lab settings; I have an older rack mount unit that is a step-ups for testing equipment at various voltages but with only a single power input available. The one I have takes 240VAC input and provides a continuously variable output in ranges from 80V to 600V single-phase.

Couple a good variac with something like a Topaz isolator/conditioner and you have exceptionally clean power. Add a reactor and/or a set of capacitors (depending on the power factor and phasing of the load) and you have seriously stable power.

 

[Chuck(G)]

When I was working stage electrics, we had motorized (Selsyn) variacs for remote control on some stage lighting.

But my point I think, is missed. There are two types of voltage regulation; line and load. A stable regulated line doesn't absolve one from implementing good load regulation. Overworked 7805s or 7805s used with insufficient line voltage "headroom" is a case in point.

Hugo, you could replace those old 7805s in the Sol supply with drop-in 3-terminal switching regulators. Hugely more efficient with better regulation--and much lower heat dissipation. Of course, this would offend the purists, but I toss it out there for consideration.

 

[Slob]

I worked for a company that was the originator of the CV and CVS (sine wave output) transformer. Ferroresonants have very good points, and some bad ones. They are very good at filtering out transients. We actually got one in that got hit by lightning (for real, not "close") and the connected equipment actually survived. However, they are not very efficient at all on less than their rated loads, and are heavy/expensive to make. They also require AC capacitors (sine wave output designs), and ever since PCB's (poly chlorinated biphenyls) were banned, it's been impossible to make a good, cheap, long-lasting AC oil-filled cap. Case in point: My mom and dad had an air conditioner unit that lasted 30 years (!) with zero maintenance. In my homes, I have had to replace the AC caps in the units every 5-7 years like clockwork.

 

[Chuck(G)]

Same situation for the start/run caps in my home's heat pump; replace them every 5 or so years. I wonder what the electrical utilities use for their PF correction caps that you see on pole-mounted installations. Those are expensive to replace.

I suspect that there are still very many old GE Pyranol caps in existence. I have a few.

I've read literature that it isn't the PCBs per se that are dangerous, but rather the product of their decomposition, PCBD (polychlorinated dibenzofuron). In any case, with the ban, it hardly matters. I suspect that a number of PCB-containing Soviet-made caps can still be had.

In the old days, one cheap way to get a high-power transformer for your amateur transmitter was to purchase a surplus "pole pig" from the local utility, dump the oil and discard the "can". You were left with a big, beefy, well-constructed core that could easy handle a "full gallon" without breaking a sweat. I think about all of that PCB-laden oil just casually discarded...

 

[Hugo Holden]

Hugo, you could replace those old 7805s in the Sol supply with drop-in 3-terminal switching regulators. Hugely more efficient with better regulation--and much lower heat dissipation. Of course, this would offend the purists, but I toss it out there for consideration.

There are a couple of reasons why changing to LDO regulators on the S-100 cards would not be idea, firstly it would ruin the originality of the card, but there is another issue.

The point at which the line voltage has just gone low enough to start to violate the required conditions for the standard 7805 regulator ( less than 7.5V at its input) is about 88 to 89V in my SOL, is not that far away from the value around 84 to 85V where ripple emerges from the SOL's own 5V discrete power regulator which powers the motherboard.

The voltage across the 54000uF filter cap on the backplane supply and the 18000uF filter cap on the Sol's 5V regulator input are driven from two different transformer secondary windings & rectifiers, that in my SOl at least are not the same, the output for the Sol's 5v system delivers 1 Volt lower on the average. So even though the Sol 5V regulator itself is an LDO design (made from discrete components), as the line voltage falls, the point where any 7805's have too low an input voltage, turns out to be very similar to when the Sol's own regulator having too low an input voltage. So there would not be much point in going to LDO regulators on the S-100 cards unless the Sol's 5v regulator was also modified to allow an even lower input voltage.

Experimenting with the constant voltage machine, I ran it all day yesterday with 7.6V on the Sol backplane, corresponding to about 90V line voltage. If it was just a manual Variac, that would be no good because at times it would fluctuate below 7.5V and require manual adjustment. But the machine kept it perfectly stable, and this give the lowest possible heat dissipation inside the computer that it it possible to achieve (keeping the computer and cards original that is).

 

[Chuck(G)]

Hugo, I said nothing about LDOs, but rather 3-terminal switching regulator modules.

 

[Hugo Holden]

Hugo, I said nothing about LDOs, but rather 3-terminal switching regulator modules.

Sorry, I was using the term "LDO" in the general sense which would include an analog or a switching type of regulator that acts like an LDO by its nature, can tolerate low input voltages and generates less heat than the original 7805 type. But in either case to get that to work would require a re-build of the SOL's own regulators as well as replacing the regulators on the S-100 cards. It would work though to get the heat dissipation down, but it would be far from original. I try to keep the computer itself as original as possible.

 

[Hugo Holden]

ever since PCB's (poly chlorinated biphenyls) were banned, it's been impossible to make a good, cheap, long-lasting AC oil-filled cap. Case in point: My mom and dad had an air conditioner unit that lasted 30 years (!) with zero maintenance. In my homes, I have had to replace the AC caps in the units every 5-7 years like clockwork.

The old oil filled capacitors for line use in ceiling fans and air con units lasted forever. But the new type progressively lose capacitance over time and fail. The films in them appear to be eaten away by voltage transients (someone called this effect "self healing") There are still some oil filled motor start and motor run capacitors available, but I think most new ones have gone to film types. A lot of small appliances now derive the low voltage supply via a X2 capacitor used as a series dropper from the line. It is good built in obsolescence as they lose capacity over a few years and the appliance fails.

Also, most single phase air con units now have gotten rid of the old AC motor and motor run capacitor and moved to inverter controlled motors. The control pcb's for these now look as complex as a laptop computer and the failure rate is much higher. In coastal areas here the fine pcb tracks corrode away or a Gecko lizard applies its waste to the pcb and its game over. I have a 20 year old air con with the old motor-capacitor system, still works. I have to repair the new units once or twice a year, faulty pipe thermistors, damaged pcb tracks from lightening hits, failing X2 capcitors, inverter IC failures and the list goes on and on.

 

[Chuck(G)]

Simple replacements e.g. for the veneered and generated 78xx are readily available from several vendors. Less headroom required, too.

Suppose you're supplying a board that draws 1A at 5V with the standard S100 8V supply. That says that pure I²R losses are 3/8 of your total power consumption. At 1A, that's 3W that might otherwise be available.

 

[Chuck(G)]

Also, most single phase air con units now have gotten rid of the old AC motor and motor run capacitor and moved to inverter controlled motors. The control pcb's for these now look as complex as a laptop computer and the failure rate is much higher. In coastal areas here the fine pcb tracks corrode away or a Gecko lizard applies its waste to the pcb and its game over. I have a 20 year old air con with the old motor-capacitor system, still works. I have to repair the new units once or twice a year, faulty pipe thermistors, damaged pcb tracks from lightening hits, failing X2 capcitors, inverter IC failures and the list goes on and on.

I'm seeing more inexpensive Chinese-origin supplies that do the single-to-3 phase conversion job, as well as offering PWM and VFD. Probably means that the old capacitor start/run setups will go the way of the fluorescent iron ballast.

 

[Hugo Holden]

I'm seeing more inexpensive Chinese-origin supplies that do the single-to-3 phase conversion job, as well as offering PWM and VFD. Probably means that the old capacitor start/run setups will go the way of the fluorescent iron ballast.

I really like the old fluoro ballast as its just a choke and practically never fails, I've got some from the 1930's era, still work fine. I cannot stand electronic ballasts, they have a limited life and generate more RFI, so despite the fact they are more efficient I'd rather have the old ones.

The Green movement do not understand that while efficiency is one thing, life of an appliance is another. When it fails it fills up our dumps with E waste and goes to recycling maybe, but the entropy, disorder and pollution of smelting more metal making more paints and plastics and electronic parts in creating more and more new appliances is much higher than if the old appliances kept working.

A 1950's vintage Westinghouse refrigerator is better for the environment as it will last for 50 years, in the meantime a new design fridge with a "superior green energy consumption rating" and a colorful green sticker on it will have to be made and trashed 5 or 10 times in that interval. What a nerve that it even has a green sticker on it.

 

[Chuck(G)]

I bought a batch of GE industrial electronic ballasts around 1995 on the surplus market. They're still in use in my workshop. I purchased them to cut down on power consumption and quiet operation; I don't detect any RFI. My garage has since gone LED, mostly for higher light output, but my workshop and office are still working on a carton of T8 lamps.

On consumer appliances, yes, I hear you. My washer/dryer was purchased in 1990 and is still perfectly functional. I've replaced small parts on them, but no real problems. I was cautioned by more than one or two service people to stay away from the current crop, unless I was willing to spring for a commercial laundromat machine.

Just boils down to design choices, mostly. Or as dad used to say, "Don't buy cheap shoes."

 

[Hugo Holden]

Attached is a photo of a commercial machine that I saw recently that does a similar job to mine, it has a motor in it that rotates the the variac's knob. I have not seen the schematic of it yet.

In my case the goal was to control the pre-regulator voltage in the computer to reduce heat dissipation. So, rather than using an RMS to DC converter IC or using the average value of the output voltage I simply generated a peak rectified and loaded voltage to resemble the electrical behavior in the computer's 8V DC supply . This also meant the computer's 8V supply could be included in the feedback loop if required.

However, if I was making a general version of the machine, I would have used an rms to DC converter IC to sample the stepped down output voltage, because the rms voltage is the important parameter and these days there can be quite a lot of distortion on the line power waveform, especially flattening of the top & bottom, as many appliances draw current on peaks.