The theory on this is correct, but the ICL7660 is not designed to provide the kind of current you'd need to run a 12W audio amplifier. It maxes out around 50mA - see figure 7 of the datasheet[1]. 50mA*17v = only 0.85W.
Because I can't see the datasheet of the audio amplifier, it may be wired in a way that it doesn't use significant current from this supply (for example, in the pre-amp part of the circuit). But then we wouldn't be getting the advantage of 17v supply to make it louder than other games.
When overloaded, simple switching power supplies like this lose their regulation. In this case, the switcher operates at 10kHZ, so it might introduce a high-pitched sound to the output.
You can probably get away with connecting the -5v input to GND to make the circuit run at 12v instead of 17v. Or, look for a more capable negative power supply.
That's true; I should've made this more clear. I'm using a "supergun" tool that is designed to lower the level back down to line-level, so I don't actually need the amplifier to run at anywhere near its full capability; in fact, that'd just be a waste. I just need to give it enough power to work at all.
Even so, I wouldn't be surprised if the high-pitched sound you can faintly hear in the video at the end is due to the effect you mentioned.
If you need more power, I'd suggest a module like this one: https://www.ti.com/lit/ds/symlink/ptn78000a.pdf
Switched capacitors are great for low currents, but inductors work better for higher currents.
JAMMA arcade PCBs normally only use the -5V for audio opamps or DACs. The power amplifier is powered by the +12V rail (in a bridge configuration to provide +/-12V swings to the speaker).
It was sort of magical when switched capacitors put all of this into a single IC. I've built a number of supplies over the years using a venerable 555 timer IC driving one side of a 1:1 transformer at about 1MHz, then rectifying it on the other side and using a linear regulator to get the "negative" voltage I want. The real trick is to understand that ground is only a concept not an absolute :-)
My favorite way of generating a negative rail using more modern components, the inverting sync buck configuration, is especially maddening. The circuit's input, the positive voltage, is the regulator's VCC: so far, so good. The output, the negative voltage, is the regulator's VEE... which sounds fine until you remember these are single supply regulators and so that's their "ground". It's the regulator output that's the mind-bending one: it gets tied to ground! A switching node output, grounded, just does not look right.
It actually makes a lot of sense once you draw it out. The startup is the truly confusing bit, but fortunately it's only unstable for your brain. These guys are actually pretty robust if you can find a part that withstands the full voltage stress (|Vin| + |Vout|), which isn't too hard these days. Highly recommended for a production-quality negative rail generator. Just don't try and do it with a regular buck; synchronous controllers only need apply.
Turns out you're right, I was working from app notes which were trying to sell sync bucks. (Not that that's hard to do, sync bucks are awesome.) I do think there are a few pathological startup cases that get easier when the output can sink current, but while those cases are easier to get into with bipolar rails, they still ought to be quite rare.
Just want to say both of these responses are awesome. Sent off a kicad file to OSHPark for a dual +/- 15V supply using the same circuit twice. Will see how it works out :-)
I'm struggling to understand one part of the article. Could anyone expert in electronics help explain this, please?
> So what if we immediately disconnected the capacitor from the rest of the universe, including its power supply and ground? Well, if we did that after the capacitor was charged, we’d find a differential voltage on the power supply. What if we hooked up ground to the side that had -5V before?
Is this a typo - should this say "to the side that had +5V before?" (ie change minus to plus)?
I think I have a handle on the design, which is that it's charging a capacitor with +5V, changing the charging side to ground, discharging the capacitor through what used to be the input (you can tell I'm not an electrical engineer, I suspect these terms are very faulty) and causing a -5V flow that way. If that's true, I think it's a typo (?); if not, and much more likely, my understanding is flawed. Insight here would be much appreciated.
First you charge the capacitor like so from the 5v supply.
+5v -> +
Cap
0v -> -
Then you disconnect the capacitor and reconnect it this way:
+5v ------> +5v
0v -> + --> 0v
Cap
- --> -5v
Compared to 0v you now have -5v. Consequently you get 10v between the two extremes. The capacitor will deplete fast. So you would repeat the cycle quickly. And add some more circuitry to smooth the output.
> Is this a typo - should this say "to the side that had +5V before?" (ie change minus to plus)?
Yes.
> I'm struggling to understand the design, but I think it's charging a capacitor with +5V, changing the charging side to ground, discharging the capacitor through what used to be the input (you can tell I'm not an electrical engineer, I suspect these terms are very faulty) and causing a -5V flow that way.
That's correct enough. It's all a matter of perspective. I'd probably say you discharge it through what used to be ground, though as the article notes technically ground is the source of electrons - but we usually talk about holes and positive (conventional) current.
Imagine measuring the voltage on a AA battery: 1.5V. Then you flip it around and measure again: -1.5V. Voltage is all relative to where you measure, usually ground. It's a tricky concept.
Maxim pioneered the charge-pump-on-a-chip concept with their groundbreaking MAX232, the first time you could get a single chip RS232 interface going without having three power supplies. Fantastic little device.
Came here to say exactly that. The MAX232 is a staple of RS232 connections to microcontrollers. +5v/-5v (maybe it goes higher?) with a couple of electrolytic caps and that's it
The dual capacitor on the output isn't really needed and it just filters out more bass sounds since essentially there are two capacitors in series.
It seems who drawn the circuit adapted it from the stereo version without much thinking.
Actually there should be no need for output capacitors at all if the internal amps are well balanced, but that should be checked with a multimeter before removing them. Good thing the Zobel network wasn't omitted, but there should also be some good capacitive decoupling on the chip supply lines to avoid motorboating and other noises.
Anyway, to get the right supply voltage the best approach is to raise +Vcc through a switching step up regulator rather than using a charge pump negative supply as they can't give the necessary current.
I probably have a few of those chips in a drawer, iirc they're used in some HAM radio equipment from the 80s, but if I had to repair a circuit using them, I'd solder a small cheap Class-D module in place of the chip since they're a lot better in pretty much everything. Probably not the preferred solution for restoring old games boards though.
Thank you! The Bose Sounddock for iPod that I found in a recycling bin uses a dual +/- 18V power supply, and I was wondering how to power it from batteries. That link taught me that the extra voltage headroom may not be required, it might run on 12V, and it would in theory be possible to build a +/- dual voltage circuit. Now I just have to decide whether it's worth building that and learning more about power electronics, or buying a new speaker and having a somewhat safer device.
In my case, I was looking for a way to get decent analog power from a USB port for a non audio related project, but a LiPo battery would present a similar challenge.
Charge pump ICs can't supply high currents on their own and their output is unregulated. It's unlikely you would get acceptable results from the speaker if you followed this blog for guidance.
Power = voltage x current = (voltage) squared / resistance.
The little rectangular 9V batteries have very high internal resistance as well as low energy capacity. The nominal 9V can easily fall below 6V with heavy loads.
If you only need 1 watt (times 2 for stereo), loud enough for average rooms with speakers of reasonable efficiency, then one 9V battery would be enough. It wouldn't last long though.
With two in series you can go louder, provided the amplifier is rated for the higher voltage.
Remember to put fuses in your power leads (except the ground return). Good quality batteries can get hot trying to power a short circuit.
It can work but the biggest problem is current capacity is generally compromised this way. The best way is to simply design the power supply right in the first place! Today that's using a buck-boost isolating switcher to create BOTH the +5V and -5V at the same time. Or a linear with the proper center-tapped transformer.
Power supply design is NOT something to leave to the last step or to give short shrift.
When I worked at Hewlett-Packard, there was a story going around that a division decided to give a summer intern the job of designing a new product's power supply. It did NOT end well - the product suffered field failures, cost a ton in warranty costs and delayed the full release of the product by a year. After that, power supply design was something that the top engineer had to be involved with especially in an intern was involved. This pairing actually resulted in both a better product and in having the intern learn more and be more likely to work for HP full time eventually.
Only for very small currents. For larger currents you use a proper DC-DC converter; another stage on your switched supply or more windings on your transformer if you're still in the Analog Age.
Because I can't see the datasheet of the audio amplifier, it may be wired in a way that it doesn't use significant current from this supply (for example, in the pre-amp part of the circuit). But then we wouldn't be getting the advantage of 17v supply to make it louder than other games.
When overloaded, simple switching power supplies like this lose their regulation. In this case, the switcher operates at 10kHZ, so it might introduce a high-pitched sound to the output.
You can probably get away with connecting the -5v input to GND to make the circuit run at 12v instead of 17v. Or, look for a more capable negative power supply.
[1] https://www.renesas.com/us/en/document/dst/icl7660-datasheet