Power Supplies

[jokkemokke]

Hello.

I recently became owner of a few Sinclair machines with which a have no experience with at all (I am normally a Commodore user. Please don't hate me). Among those machines there is a ZX Spectrum and a ZX Spectrum+. The machines only came with one power supply: a Sinclair UK 1400 with DC barrel plug.

Before I connected anything I decided to measure the output of this supply, and it measured 12V DC with positive in the center and negative on the rim. The machines are marked with 9V DC, but I read on this forum that it is not unusual that the Sincalir supplies measures 12V when they are unloaded.

But what really confuses me is the polarity of this supply. Shouldn't both of the machines have negative in the center and positive on the rim? At least that was what I could find from googling around.

It seems like a strange defect on the power supply. I mean why would it still work, but suddenly change polarity?

Can anyone clarify this mystery?

[Ast A. Moore]

Yes, 12V and even more is typical for a Spectrum PSU without a load. However, the swapped polarity is definitely an oddity. Are you sure you haven’t accidentally swapped the leads on your multimeter? Can you double-check with, say, another known-good PSU?

I recommend getting a modern switch-mode, regulated 9V 2A (or higher) PSU and use that instead of the original one with your Spectrums, anyway. The internal voltage regulator won’t have to dissipate all the extra energy as heat as much, and the machines will run so much cooler.

[jokkemokke]

I already double checked the leads of the multimeter to make sure that I was not crazy

Maybe an explanation could be that one of the previous owners has fiddled around with it. I will open it one of the days to check it inside.

I will see if I can find an universal PSU that can deliver 2A. Probably a good idea anyways, since there is only one supply for the two machines.

[Seven.FFF]

This one works well. Just make sure you get the centre-negative version, because northpada also make a centre-positive model.

https://www.amazon.co.uk/dp/B072PRLV26

Byte Delight also sell a good one.

[jokkemokke]

Thanks, I will have a look at that one.

I was just thinking if it would be possible to just make an adapter cable for a PC power supply. I have plenty of those lying around and they should be able to give sufficient current.

Would it be a problem to supply the machine with 5V instead of 9V? Or maybe better with 7V, if I use 12V minus 5V. I guess there is a probably a LM7805 or similar inside the machine, transferring the voltage down to 5V anyways. Would that work?

[Ast A. Moore]

Would it be a problem to supply the machine with 5V instead of 9V? Or maybe better with 7V, if I use 12V minus 5V. I guess there is a probably a LM7805 or similar inside the machine, transferring the voltage down to 5V anyways. Would that work?

1. Most LM 7805 voltage regulators will require at least 7 or 8 volts of input voltage to operate reliably. I’d say feeding it 7 volts is really pushing it.
2. The 48K/128K Spectrum feeds the unregulated +9V from the PSU directly to Pin 4 (lower side) of the edge connector, and some peripherals expect it to be there. If you’re only supplying 7V, even if the regulator works, you’ll end up running into potential issues with peripherals.

In short—not a good idea.

[1024MAK]

The Sinclair ZX Spectrum 16k, 48k and Spectrum+ machines use DRAM chips that need three DC supplies:

• +5V
• +12V
• -5V

The video circuitry also needs a +12V supply.
The rest of the machine runs off the +5V supply.

So how did Sinclair do it, given that the external power supply unit (PSU) is only rated for a nominal +9V?

The answer is that for the +5V, they used a standard series pass element 7805 voltage regulator IC. This requires at least 7V or more. It's best to supply it with a minimum of 7.5V so that it has a bit of headroom to allow for voltage drop in the cable and the power connector.

For the +12V and the -5V, Sinclair designed a DC to DC converter/inverter circuit. This takes the nominal +9V and produces the required outputs. The recommended input range is between 8V and 11.9V (on-load voltages).

Later issue machines can work with higher input voltages. As the a DC to DC converter/inverter circuit was revised to work correctly with higher input voltages.

Now if you have an external PSU that can supply +5V, +12V and -5V, you can ignore the Sinclair 9V PSU. But as pointed out above, some expansions use the +9V edge-connector supply. Some manufacturers used the +9V to supply extra 7805 ICs in the expansions. Others use it in other ways.

Mark

[jokkemokke]

Thanks for the explanations.

Just an update from my side. I opened the Sinclair supply, and yes; it looked like someone had been doing some repair work on it and for some reason swapped the polarity. I swapped the cable back to normal polarity, plugged it in to the machine and it worked

One of the machines seemed to work without problems, the other one seemed to have maybe some RAM issues (garbled screen on startup). But I guess this be a good fixing/debugging project for the winter.

[silencer82]

Hello everybody,

I am working on a project with a 48k Spectrum
which I have found in trash. Its case is almost completely
destroyed by previous owner but the computer works fine.
I am trying to make a laptop with this mainboard and a small LCD
screen that is using 12V as power supply. I have an old laptop housing that has a brand new big battery. The battery provides 5V and 12V on its output port. Would it be a big problem if the Spectrum ran on 12V instead of 9V?
I am building this just for fun.
Thank you for the help.

[Mad Fritz]

The 9-ish Volt supplied to the Spectrum are AC, not DC, and converted to +5V, -5V and +12V internally IIRC.

BR / Andy

[Ast A. Moore]

The battery provides 5V and 12V on its output port. Would it be a big problem if the Spectrum ran on 12V instead of 9V?

There are drop-in switching replacements for the LM7805 voltage regulator, such as this one. If you’re planning to use an external power supply with a 12V output, you should consider replacing it to eliminate the extra heat the original LM7805 will generate to dissipate all that extra energy.

Keep in mind, though, that if you’re planning to use the edge connector of the Spectrum, it’s best to proceed with caution when using peripherals that obtain their power from it. Pin 4A just passes through the positive voltage from the power supply directly. The Spectrum’s original PSU will drop its output voltage down to about 9V under load. An external switching PSU will easily maintain its nominal voltage for up to its rated power output. I suppose, many peripherals will be able to deal with it, but presumably not all.

Also, the 9V rail is used to energize the built-in speaker via the collector of TR7. Additionally, it’s used as a reference point for generating negative voltages within the Speccy. I don’t know if it’s a good idea to exceed it by as much as 3V (9V vs 12V).

[Ast A. Moore]

The 9-ish Volt supplied to the Spectrum are AC, not DC.

Incorrect. It is DC.

[1024MAK]

Keep in mind, though, that if you’re planning to use the edge connector of the Spectrum, it’s best to proceed with caution when using peripherals that obtain their power from it. Pin 4A just passes through the positive voltage from the power supply directly. The Spectrum’s original PSU will drop its output voltage down to about 9V under load.

The output of the Sinclair ZX Spectrum external PSU depends on the output current. It can be around 10.5V to 11.9V with just a ZX Spectrum connected. When loaded at its rated output current (1.4A) its output should not drop below 9V.

I do agree that using a switching regulator / switching DC/DC converter in place of the 7805 is a wise choice.

Also, the 9V rail is used to energize the built-in speaker via the collector of TR7. Additionally, it’s used as a reference point for generating negative voltages within the Speccy. I don’t know if it’s a good idea to exceed it by as much as 3V (9V vs 12V).

Yes the speaker is driven from the (nominal) +9V rail. But the voltage on this rail will not affect the speaker, as it is fed via a NPN transistor in an emitter follower circuit. The voltage at the emitter cannot (in normal operation) go above the base voltage. However, if the software crashes (or is badly programmed) and the transistor is left switched on, any extra voltage on the +9V rail will increase the power that the transistor has to convert to heat. If the +9V rail is actually at 12V, and the transistor is left switched on, I estimate that it will be running at about 0.7W. It is rated at 1W at an air temperature of 25°C. But keep in mind that the temperature inside a ZX Spectrum (or any enclosed space) may not stay at 25°C and could climb to 40°C or more. If the transistor overheats, it may be damaged.

The +9V supply itself is not used as a reference voltage by any circuitry inside the machine. The +5V supply from the 5V voltage regulator (7805 or any equivalent switching regulator or DC/DC converter) IS used as a reference to generate the correct voltage for the +12V supply (from the internal DC/DC converter formed by TR4, TR5, the “coil” and the other components in this area.

However if the +9V supply is greater than 11.9V, the regulation circuit for the DC/DC converter circuitry (TR4, TR5, the “coil” etc..) may not work correctly. Please read on...

What IS VERY important, is that the input voltage on the +9V rail MUST AT OR BELOW 11.9V unless the board is using the latest modification (see the details here). Issue 5 and issue 6a boards already have this design. Issue 4a and 4b have a variation but may not have the latest version so need to be checked. I have not checked these myself and I have not checked an issue 4S yet (believed to be similar to a 4a). Earlier issue boards are unlikely to have been modified unless they were returned for service/repair. However any board between issue 2 and issue 4a/4b/4S May have already been modified.

If a board does not have the latest version of the DC/DC circuitry, either modify it so that it is brought up to the latest modification version or don’t connect a supply of greater than 11.9V. Connecting a supply of more than 11.9V to a board that is not to the latest modification version may result in the -5V rail being out of specification, which may result in one or more of the 4116 DRAM chips from malfunctioning with possible permanent damage.

The reason that the problem exists in the earlier design of DC/DC converter circuitry, is that there is a DC path from the +9V rail to the +12V rail via the “coil”. When the +12V is above +12V, TR5 will start to switch off causing TR4 to reduce its output power, which in turn reduces the AC voltage produced by the “coil”. The system thinks that the voltage on the +12V rail is too high, so it is trying to reduce the voltage on the +12V rail. But the DC can flow via the coil from the +9V rail to the +12V rail. With a lower AC current in the coil it’s AC output is low or absent, so the system is unable to produce a -5V rail.

Mark

[Mad Fritz]

The 9-ish Volt supplied to the Spectrum are AC, not DC.

Incorrect. It is DC.

Of cause right, too much booze I suppose