Eskom voltage range

What’s the expected voltage range for the grid in SA? Is 245v normal or too high?

I’m helping some family members out with “tech support” and they told me about an issue they had recently.

They are dentists, so not very technically inclined.

They recently constructed new consulting rooms and installed a new German x-ray machine. They also installed an 8Kw Deye inverter, PV panels and enough batteries to make it through loadshedding.

Unfortunately, the X-ray machine blew during loadshedding – apparently due to too high voltage. In other words, supposedly while being supplied from the inverter.

The X-ray company sent a guy who measured the supply voltage at the DB (e.g. inverter output, but not sure if Eskom was available at this point) and he claimed it was 249v and that high voltage caused the damage. I went and measured it myself at a plugpoint on the Eskom side and found it 245v.

They changed inverter settings to lower the output voltage but that resulted in the inverter disconnecting from the grid - probably due to being incompatible and disconnecting itself. And since there was very little sun the last couple of days in Pretoria, this resulting in them running out if battery power.

They ordered a PSS TND 2kVA voltage stabilizer to basically plug in between die inverter and the X-ray machine, and will then switch the inverter back to the previous settings and let it go back to the Eskom grid levels.

Is that standard? Anything else they can do?

You should get an output voltage reading from the Deye inverter. I doubt it is the inverter doing too high voltage. Eskom, yes.

This is from my Goodwe inverter. You can clearly see where the inverter provided around ~231v average during loadshedding and when online with eskom, it follows the eskom voltage/frequency and its much higher.

I doubt you can blame voltage alone.

Here is my grid voltage… for the 2 hours today that there was actually grid available in Pta. Highest peak was 238.6V. I agree that the problem is Eskom not the inverter. You can actually setup the inverter to disconnect from the grid if the voltage is too high.

My suggestion would be to charge more aggressively from the grid when it is cloudy/rainy and setup the parameters so that the inverter carries the load where the grid is not within spec of the x-ray machine

Phil shared a nice post on the grid code values before

So 245V is still in range

230V plus or minus 10% according to SANS 10142 (5.3.2). So up to 253V and as low as 207V. So 245V is not too high.

Interestingly, I see Australia has 10% room at the top, but only 6% at the bottom, so 216V minimum. Even in SA, I think we’ll consider anything below 220V a reason for concern, much more so than we’d worry about being above 240V.

What would happen if the X-ray machine gets too low a voltage? Wouldn’t that increase the current and also damage it?

I’m not sure what happens in an X-ray machine, and what the start up needs, or instantaneous needs are when an X-ray is taken. Transient spikes in power draw could be high? :man_shrugging: Guess the inverter should disconnect itself when the output voltage drops too much… or it didn’t drop enough to disconnect?

Don’t forget to check for faulty inverter to. The house opposite from us installed n sunsynk unit about 3 weeks ago, first night they had problems. 5 days of sparkies coming and going, one fridge, microwave, garage motor, light bulbs and garden water computer burned out, they pinpointed that it is a bad inverter and replaced it, problem solved.

if it was in the middle of active load shedding the general Eskom voltage should not play a role? But there does seem to be increasing reports of stuff blowing due to high voltage as the grid is connected again. If it ispossible to set a delay before the inverter re-connects to the grid that could maybe help decrease risk of passing that high voltage to the connected equipment.

Is the one minute reconnection requirement not already part of the grid code?

A friend of mine recently told me his inverter popped (Voltronics type). This made me think: I am always quite careful when selecting a PSU for the computers I build as a bad PSU can take your other parts with it when it goes. What happens when an inverter pops? Is there protection built in? @plonkster does Victron have anything specific worth mentioning? Are there tests one can do to early identify an inverter that is on its way out?

There is something else to consider as well, and that is that some kinds of loads (usually those with very poor power factor, and a non-linear current waveform) can actually cause an inverter to overvolt. To use a very simply example, if you take one of the small Victron Phoenix inverters, and put a camping hair-dryer on it on half-heat (ie, it switches a diode into the line so only half the AC waveform is used), the inverter will actually switch itself off on an overvolt condition (to protect the loads).

See also this:

What I am saying is… it is important to match your loads with the inverter tech you are installing. And this is another place where an old school LF design really makes a lot of sense. Now of course I realise that the Phoenix I referenced earlier is also an LF design, but in that case the inverter was very evenly matched with the maximum power of the appliance. For sensitive things, you may well do better with an oversized LF setup.

Recently learned that one test for an inverter under repair is curlers/heat guns, due to how they work.

NRS 097-2-1 refers to “not before 60 seconds”

see also

1 Like

Yes, but what I mean is that there’s already time for the grid to stabilise in that 1 minute. Do you need more?

In that case I’m pretty sure that it’s the impedance of the LF transformer that caused the issue - the voltage sagged on the half-wave under load, causing the control loop to compensate, which caused the unloaded half-wave to go overvoltage.

I’m beginning to think some odd number slightly higher than 60 might be a good idea. Remember, ALL the NRS097 inverters reconnect at the same time if you use the default 60 seconds. Maybe use something like 103 seconds. No, don’t use that… I’m going to use that. You get your own number.


That’s probably because they are high loads in compact form. Heat-guns usually don’t use the diode-method, they have an element arrangement that allows different power levels.

It should be a random value, between 120 and 900 seconds IMO.

1 Like

If I know my local voltage is especially temperamental (think the Gauteng people might have this worse) I would opt for a longer delay if I have the option

Wouldn’t all those inverters reconnecting cause voltage drop rather than spike (recharge batteries)? PV may only reconnect slowly over 10 minutes?