ZX81 with 1080p60 HDMI output

[Flatulentia]

Hi peeps,

I've just spent the afternoon getting one of those cheap 'MINI AV2HDMI' branded HDMI converters working properly inside my ZX81.

The output isn't perfect, but it is almost spotlessly clean, so it's been well worth the effort in my opinion as I now have a ZX81 I can use on any modern display rather than having to hope that one has a UHF TV input then putting up with a lousy picture anyway. None of my displays have a composite video input, so that wasn't an option either.


There are a few caveats to getting one of these working well internally, so I thought I'd list them here:

- Use a back porch generator if you have an early ULA. Anything based around Zigg's design should be fine.

- Use proper 75 Ohm braided video coax to connect your composite video signal to the converter to avoid stray pickup of digital noise.

- Solder a link between both audio inputs and ground to ensure silence on the HDMI audio channels. You may get minor digital noise breakthrough on the HDMI audio if you don't do this.

- Decouple the converter's power supply with a ceramic capacitor (1uF works fine) to suppress digital noise on the +5V rail. This is especially important if you've replaced the ZX81's original 7805 linear regulator with a switch-mode one. I'd recommend this anyway as it reduces ripple current in the ULA.

- Fitting one tidily inside the ZX81 case is impossible without removing the converter's case and RCA input jacks first due to a lack of space. The jacks can be removed by carefully nibbling away at them with a pair of electrical sidecutters if you don't have access to a desoldering station.


I've yet to modify the case to give access to the HDMI port from the outside world but will post photos soon showing everything in-situ.

Total cost was under £16 including £10 for a ready-built Zigg's back porch mod and £4 for the HDMI converter from eBay, and I'd say it's been well worth every penny.

EDIT: Photos added. I was hoping to show off some nice graphics, but my keyboard tails disintegrated when I was putting it back together. A new (2018-ish) keyboard has been ordered and should be arriving in the next few days.

The magic black box in the photos is where I've housed all of my extra electronics. The original UHF modulator is stored away safely in my parts drawer. They're already worth fairly serious money on eBay, so it seemed worth holding on to it in its original unmodified form.

Mounting the converter with the HDMI port upside down was by far the easiest option and, thankfully, the picture still comes out the right way up.

[Lardo Boffin]

Sounds interesting!

Can you add some internal photos?

[Flatulentia]

Sure. It'll have to wait until daytime tomorrow (UK) though so I have plenty of light to get good photos. I should have it properly mounted by then with a hole in the case for the HDMI port.

[1024MAK]

- Decouple the converter's power supply with a ceramic capacitor (1uF works fine) to suppress digital noise on the +5V rail. This is especially important if you've replaced the ZX81's original 7805 linear regulator with a switch-mode one. I'd recommend this anyway as it reduces ripple current in the ULA.

Ripple current in the ULA?

Do you mean the noise on the +5V rail? Only with either a 7805 or a DC/DC converter/switching regulator there should be no discernible ripple on the +5V supply…

And the ULA internal logic matrix cells operate off an internally regulated supply anyway.

However, the noise (digital switching noise) from the logic chips may be at a high enough level to affect the analogue video, especially if being converted to HDMI resolutions.

Mark

[Flatulentia]

My oscilloscope is out on loan at the moment, so I can't measure anything accurately other than DC voltages and low frequency AC.

All I know for sure is that the same HDMI converter was providing reasonably clean output with the original 7805 but became almost unusable when I swapped the regulator out for a switch-mode equivalent. Adding a 1uF ceramic capacitor directly across the output pins of the switch-mode regulator produced a far cleaner result than the 7805 without one. If I was to put the 7805 back in, I suspect that the same capacitor would have helped that too.

The 7805 is far from ideal as its output impedance rises with frequency, so I'm not convinced that the four 47nF capacitors remotely splashed around the board with long tracks in series (effectively behaving like inductors) are sufficient on their own. Datasheets have always stated that a 7805 needs a 100nF capacitor (ideally ceramic) directly across its output pins which the ZX81 designers chose to ignore making it farther from ideal than it could have been.

Looking at some of the photos around here, I'm not using the same potted switch-mode regulator as many of you, so maybe that's where the problem lies. I suspect that the switching oscillator frequency is poorly filtered on mine so is appearing at the output as HF ripple, but I won't know for sure until I get my oscilloscope back. As we have no control over which switch-mode regulator someone else may be using, I still advocate fitting the capacitor.

I've not had much experience with TTL logic, but a company I used to test for had close to a million CMOS logic ICs pass through their doors while I was there and even those were prone to early failure due to HF noise on the supply rails despite their wide operating voltage range. It took me months to convince the design engineer to decouple everything properly. Once he'd eventually listened, failure rates of new units in the field fell dramatically.

The knock-on effect was that 2 of the 3 repair engineers lost their jobs a couple of years later (actually repositioned as test engineers, so I didn't feel too bad about it) as hardly anything ever came back for repair under warranty after that.

If this doesn't also apply to TTL logic in general then I'm happy to be corrected, but bear in mind that many of us are upgrading to a CMOS Z80, so a little extra caution only costs pennies and certainly won't do any harm.

I had no idea that the ULA had internal regulation. Interesting.

[1024MAK]

Ah, your suspecting the leakage of the switching frequency from the DC/DC converter/switching regulator. Or the potential unstability/self oscillation of a series pass regulator (7805 in this case).

Whereas by ripple I was thinking that you were meaning the 100Hz ripple from the unregulated 9V DC supply.

Yes, as with many computer designs from the 1980s, especially consumer market low cost designs, sometimes the amount and values of the capacitors used for decoupling the power rails is, shall we say, minimal. And Sinclair’s aim was to produce the lowest cost home computer.

74xxx and 74LSxxx logic chips and the (typically) NMOS microprocessors and memory chips available in the late 1970s to mid 1980s are relatively forgiving and may work with minimal power rail decoupling.

More modern devices, especially fast CMOS technology, definitely needs good decoupling.

Anyway, as seen in this topic “Pictures of my ZX81 fitted with a RECOM 5V Regulator”, I do recommend fitting two extra 100nF ceramic capacitors, both very close to the regulator. One across the input (9V nominal) supply and one across the output (+5V/VCC).

There are more Ferranti ULA details in these posts/topics:
https://www.sinclairzxworld.com/viewtop ... 566#p22566

https://www.sinclairzxworld.com/viewtop ... f=7&t=4185

I think there is even more if you can find it… (I’m sure either Moggy or myself posted some other Ferranti ULA datasheets).

Mark

[Flatulentia]

There are more Ferranti ULA details in these posts/topics:
https://www.sinclairzxworld.com/viewtop ... 566#p22566

https://www.sinclairzxworld.com/viewtop ... f=7&t=4185

Good stuff! Thanks for sharing.

[Flatulentia]

Photos added to first post.