The great MSX power consumption thread!

Pagina 1/2
| 2

Door RetroTechie

Paragon (1563)

afbeelding van RetroTechie

25-05-2019, 09:25

Calling out for your help! To gather something that's currently missing in the 'MSX universe': a kind of database / overview of the power consumption of various MSX models. Especially: the ACTUAL power consumption in use, how that power is divided over various DC voltages, and peak vs. normal currents. NOT so much AC that's pulled from the wall, and even less what's written on the type label (although both figures can be useful). Afaik, such a 'database' does not exist.

Why? Mostly this: as you may have noticed, there's regular threads about electrolytic caps dying, internals of some MSX models getting rather hot, 100 <-> 120V issues, some external 'brick' missing, etc. And then usually follows some discussion about what could be used as replacement, how 'heavy' diodes are needed, what voltage(s) or current(s) a replacement brick should be able to supply, etc.

But the most important info is usually missing: what the machine in question actually consumes under normal use. Service manuals, type labels, component datasheets etc, or an educated guess is useful but a poor substitute for this info. Having a clearer picture on this would make it MUCH easier to recommend specific solutions, replacement psu models, or eg. what voltage regulator(s) would be up to the job.

What am I looking for? Primarily: actual MEASUREMENTS of the power consumption of MSX machines, under 'normal' use. Ideally of the DC currents consumed from the internal +5V, +12V and (rarely) -12V. With as few as possible cartridges inserted. If possible: peak, normal & lowest values seen.

Special points of interest:

  • Info about how posted figures were obtained (measurement method).
  • What hardware configuration was tested (for example: what cartridges were inserted if any. Internal ROMs replaced with EPROMs. NMOS vs CMOS Z80s. TMS9918 vs that Toshiba VDP, etc).
  • UNregulated DC voltages found in specific power supplies (to calculate heat dissipation for their voltage regulators).
  • The power consumption ADDED by various types of floppy drives (both in rest & during seek or read/write).
  • The effect of what software is running (if any).
  • The effect of speed upgrades like 6 or 7 MHz circuits.

If you own a Kill-A-Watt or similar device, and an MSX computer, then you can help! Actual power consumed (turned into heat inside your MSX) not "reactive power" is what's wanted. Although power factor, and efficiency of some MSX psu's might be interesting.
If you are a hardware person like myself, know how to measure voltage or DC current, and don't mind pulling the top of your MSX machine to take a few measurements, then you can help!
If you know such a person & can bring him/her and your MSX machine together to take some measurements, you can help!
If you have an MSX model for which no picture of the type label (with voltage / watts info) can be found on the internet, or an overview of what parts are on the mainboard, then you can help!

What am I NOT looking for?

  • Info on a type label that's easily found on the internet.
  • A power consumption figure that you found in some floppy drive's datasheet.
  • Power consumption figures for individual IC's found in MSX machines.
  • Less-accurate or less-complete figures for a model whose 'twin sister' has already been covered. But in the case of good measurements, multiple data points don't hurt.
  • For the most part, power consumption of your run-of-the-mill MegaROM, memory mapper or SD card reader. Cartridges are external, extra, I can take measurements in that area myself, and regardless: there's a spec about what cartridges are allowed to consume. If a cartridge goes over that because designer didn't do his/her job, not my problem.
  • Likewise for educated GUESSES concerning MSX model X, Y or Z.
  • Lots of discussion about related subjects that 'pollutes' the informational contents of this thread.

Why am I asking YOU for help? I know how to pop the hood of an MSX machine, take some measurements & post results here. If you'd lock me into a room filled with MSX machines with the necessary tools, I could gather figures quickly. Buuuutttt... I only have a small collection of MSX machines myself. Owners of some unique MSX models are few & far between. And most of all: many hands make light work. If done correctly, a single measurement session is all that's needed for each MSX model, ever. Including machines that have identical hardware.

Consider it like writing some notes into a service manual. If you know the data for a Philips VG8020/20, then you know it for a VG8020/40, since those machines have near identical hardware inside (not the VG8020/00, though!). If you know it for a Sony HB-F700S, then it'll be the same for an F700P. Many examples like that.

I hope this can be a long-running thread, where from time to time someone (including myself) will post some figures measured for MSX model X, Y or Z that wasn't yet covered, or in more detail. Along the lines of that 'look under the hood' thread a while ago. Once enough figures are available, I intend to compile an easy-to-search list / spreadsheet or similar. For questions about how to measure things, or what readings mean: just ask.

Happy hunting!

Aangemeld of registreer om reacties te plaatsen

Van gdx

Prophet (3083)

afbeelding van gdx

25-05-2019, 10:59

Consumption is often written under the machine, under the power supply or in the manual. This information is also indicated as much as possible in our Wiki.

Van Meits

Scribe (5644)

afbeelding van Meits

25-05-2019, 11:32

@RetroTechie:
I'd love to help you on the data from my computers and I'm fairly certain I have the necessary tools. However I doubt I have enough knowledge about the subject.
I do have several MSXes, a basic multi meter, a KWH meter for between mains socket and power line, a cartridge with a cable and probe points on all pins without the chance of short circuiting.
The problem is that except continuity, voltage/resistance measurement and diode testing only to find the + leg of a LED I know nothing about my multi meter.
I'd expect you to rather want an as much as possible base figure, which in my book would mean that there's nothing in the cartridge slots.

I'll have a tech read your post and help me perform the tests.

@gdx:
That's not the info he wants. A PSU is always supposed to be (a bit) stronger than required/used.

Van RetroTechie

Paragon (1563)

afbeelding van RetroTechie

25-05-2019, 13:39

Meits wrote:

@gdx:
That's not the info he wants. A PSU is always supposed to be (a bit) stronger than required/used.

And not just that. I'd like for us to have some figures where the efficiency of whatever psu is in a machine, is NOT part of the equation. Let me give an example:

Let's say there is a hypothetical MSX machine that only uses +5V internally, doesn't do +12 or -12V at all (okay this violates MSX standard), and has a power supply that consists of a mains voltage -> low voltage AC transformer, bridge rectifier, capacitor, and a 7805 style linear regulator. Let's say a measurement shows the mainboard in this machine draws 1.4A from that +5V (= 7W). And for simplicity sake, let's assume this is very constant, not depending on what software is running on the machine.

I'm interested in that "1.4A on the +5V line" figure for this machine. For clarity: that info can NOT be found in any existing documentation that I'm aware off. Not for any MSX model. So for people contributing data to this thread, could be considered research into unknown MSX territory.

Now let's suppose that mains transformer puts out some 8V AC, and after rectifier + capacitor some 9.5V DC is left. This could be measured with a voltmeter on the input pin of the linear regulator. Again this 9.5V figure is nowhere to be found in existing documentation. Note that this figure varies with the mains AC voltage going in. So for anyone doing that measurement, MEASURING the mains voltage going in would be useful to note.

With that info, one can calculate that the linear regulator dissipates roughly (9.5V - 5V) x 1.4A = 6.3W. That number is mostly interesting to get an idea how much heat is shed in the heatsinks of various MSX machines.

Now let's suppose that overall, the mains AC -> +5V conversion is 40% efficient (it couldn't be higher than 53% in this example anyway). That means to provide 7W on the 5V rail, power supply draws 17.5W from mains AC. The remainder (17.5W - 6.3W - 7W = 4.2W) is various losses in mains AC filter components, fuses, transformer, bridge rectifier, and perhaps even capacitor.

Now engineering + marketing department may have done similar measurements one day, and decided to write "230V, 50/60 Hz, 20W" on the type label. With the above example, can you understand (a) how little that number says about power draw for the mainboard, and b) how useless it would be to decide what you'd need to replace the internal power supply?

Van Meits

Scribe (5644)

afbeelding van Meits

25-05-2019, 14:17

I've checked the tech guy. I'd need an amp clamp and a place where the 5V is "clampable". Unfortunately I do not have such a device.
I can measure the watts all my machines draw if you wish, but that's all :/

Van thalin

Rookie (19)

afbeelding van thalin

25-05-2019, 21:04

The original SVI PSU for SVI-728. Is rated
9V 1.5A
16V 0.8A

My measurements on the AC input (no Cartridge or any other devices connected)
~10Vrms@1500mA
~18Vrms@160mA

Van gdx

Prophet (3083)

afbeelding van gdx

26-05-2019, 12:45

Only the indication given by the manufacturer is relevant.

Self-measurement of intensity and tension will only tell us whether our power supply works well or not.

In general, the voltage measured at the output of the old power supplies is higher than indicated because it drops according to the consumption.

The voltage is calculated according to the assumed consumption range.

The intensity (or power) indicated on the power supplies is the maximum current (or power) that it can provide.

The power (or intensity) indicated on the machine is the maximum current (or power) that it can consume (or necessary).

So if you want to make a power supply, it must be able to provide a current at least a little more than what is indicated on the machine or at least the same as what is indicated on the power supply.

You must also take into account the voltage drop depending on the consumption if the power supply is not regulated precisely.

Van RetroTechie

Paragon (1563)

afbeelding van RetroTechie

26-05-2019, 14:09

Thanks for that, thalin! Smile2 I happen to own a SVI-728 myself, but without the AC adapter. Iirc it has some stupid power supply with video circuitry mixed on the same board. And using transistors rather than IC regulators. Very unusual, more trouble to feed in external DC power if desired. But mine has just that, so will 'dig up' that machine & post some more accurate numbers when I find the time.

@gdx: have you read my previous post(s)? This thread concerns what goes from INTERNAL power supply to mainboard (+ analog board, floppy drives etc). That is: the DC currents on +5V, +12V and -12V lines. NOT AC power pulled from the wall.

Van gdx

Prophet (3083)

afbeelding van gdx

26-05-2019, 15:39

RetroTechie wrote:

have you read my previous post(s)? This thread concerns what goes from INTERNAL power supply to mainboard (+ analog board, floppy drives etc). That is: the DC currents on +5V, +12V and -12V lines. NOT AC power pulled from the wall.

Me too I don't speak about AC power pulled from the wall. I speak about AC and DC currents output. We can deduct DC currents output with reference on chips for DC and AC output can be deduced with it if nothing is indicated on the transformer.

AC adapter of SVI-728 provides two alternative voltages: 16V (0.8A) and 9V (1.5A) but I don't know the pinout of the connector.


The AC to DC circuit is in the "Service and Technical Manual - SVI-728 Computer System" pdf but the pinout is not described.

Van thalin

Rookie (19)

afbeelding van thalin

28-05-2019, 10:50

The output is also higher since it was designed for 220V but most countries sre now using 230V.

Van gdx

Prophet (3083)

afbeelding van gdx

28-05-2019, 12:06

The devices are designed to support a tension of +/-10%.

Pagina 1/2
| 2