ZX Spectrum Z80 Keeps Resetting
thorh at ismennt.is
Tue Jan 2 10:58:51 CST 2018
The power supply voltage might be a bit too high.
Early Spectrums expected a "9V" DC power supply.
The DC/DC converter that makes +12V, +12VA, -5V and -12V(AC) in some early
versions (up to Issue 3, IIRC) stops working if the input voltage is too high.
Happened frequently around midnight here in Iceland
when one had typed in (but not saved) a long program,
because normal people then turned off their lights,
and the mains voltage went a bit up :)
----- Original Message -----
From: "Rob Jarratt via cctalk" <cctalk at classiccmp.org>
To: "General Discussion: On-Topic and Off-Topic Posts" <cctalk at classiccmp.org>
Sent: Monday, January 1, 2018 11:46:13 AM
Subject: RE: ZX Spectrum Z80 Keeps Resetting
Replying to my own mail to consolidate my answers to the two very kind
responses I got.
In answer to Adrian: Regarding the PSU, I actually have two Spectrums, the
same PSU seems to power the other one OK. I quickly checked it and it is
outputting 13.4V and there is no ripple to speak of. So I think the PSU is
In answer to Jon: I did look at the power rails. The output from the 7805
looks absolutely fine and the inputs to some of the ICs looks fine. However
the Vcc input to the Z80 did look a bit noisy, I found there are quite a few
spikes, their amplitude appears to be 600mV. I temporarily added a 3.5uF
capacitor I happened to have lying around, this reduced the amplitude of the
spikes to about 200mV, but didn't affect the behaviour. I am not sure if
these spikes could cause the reset behaviour though. I suppose the spikes
could mean either there is a faulty IC (finding that won't be easy), or
there is a bad capacitor somewhere. I did replace most of the electrolytic
ones, but not all of them, so that is probably a good line of inquiry.
I don't think it will be a bad memory location/region in the ROM though
because a lot of the resets occur in a loop, so it can read the locations,
although I suppose it is possible that the logic levels on the address/data
paths could be marginal and occasionally resulting in bad data. My next step
was going to be to discover how to get my logic analyser to capture the
addresses *and* the resulting data, but I think I will double check the
Happy New Year!
More information about the cctalk