Which brands, in general, preserved the true black reference in picture, either by DC-coupled video, or the use of a DC-restorer circuit?

Not considering the smaller local makers and assuming we are talking about early B&W sets, these are some that I can think of that generally made some effort to retain black level:

Astor
Atlas
Ekco
HMV
Kriesler
Philips
Precedent
Pye

The ones that did not, at least in the early days, were those that followed then-current US practice:

Admiral
AWA
Healing

There are exceptions, such as the last B&W hybrid chassis Thorn (shared with AWA) which has AC coupled video.
This model was generally regarded as rubbish!

That's an interesting list. You would think AWA would know better as true black reference is a matter of picture fidelity, but they chose to dwell in the low end along with the likes of Admiral and Healing! Although I would rank Admiral above Healing engineering-wise. Healing also had slight video smear whereas HMV avoided this with rounded I.F. response shape (based on British engineering experience?) Thorn's long held AC coupled video philosophy (the odd man out in UK?) dates back to 405 line broadcasts where 16mm programme sources often had crushed blacks and AC coupling, it was judged, allowed dark scenes to automatically lighten up for home viewers.

AWA did change their ways eventually. I'd have to scan through the schematics to find out when. Often in these cases it's a matter of an engineer retiring, dying or moving elsewhere!

I think I might have included a post of a screen shot from an HMV E1 - 01 showing a PM5544 test pattern and resolving the 4MHz and 5MHz grates in my post on the modulator I designed that strips out the rubbish in the VBI. It's hard to believe that's going via a modulator and the TV's RF but it sure is.

CCIR being so different to the British 405 line system, I suspect the local engineers looked more to European designs (especially German) for inspiration, not to the UK so much. And of course they were the only Oz maker to use the RCA Synchrolock circuit (in the E and F series) well after RCA had decided it was too expensive! They continued to use it in high-end "fringe area" chassis well into the 60s because it worked so well with a noisy signal.

There was a bit of UK tech in the E1 but it was that part of it that almost ruined the company. Most E1s had to be extensively modified, UK-sourced CRTs and Horizontal Output transformers (flybacks) mostly all failed under warranty. The following F series learned a lot of hard lessons from the E series.

What of the TVs that had a .1μF cap from video output to CRT but there was also a 100k resistor across the cap: Does that maintain DC reference? Always wondered about that circuit feature.

That cap and resistor provides beam current limiting by biassing the CRT back as bean current increased. At higher brightness and contrast settings it also has the effect of reducing the amount of DC restoration. Debates used to rage about the desirability or otherwise of 100% DC restoration and many designs used this approach to have a bet each way.

Another reason it was done was to avoid excessive beam current loading the (generally unregulated) horizontal output and causing large changes in width and height.

That explanation makes sense.

Just having a look at some early AWA circuits. They seem to be partially DC coupled, C230 (220n) in parallel with R230 (330k) in their first chassis for instance, as per Ian Robertson's explanation. From model 207-C onwards; AC coupled, but then from D1 and D3 onwards, DC coupled! AWA originally followed RCA designs.

There in an article in R, TV & H which defends AC coupling, as their TV designs were AC coupled. It points out that AC coupling was the norm in the USA at the time.

RTV&H attitude was the TV stations didn't do a good enough job maintaining accurate black level and to a certain extent in some cases that was true. - ATN7 was worst offender.

I guess it was all too hard, it would have been so difficult using the technology of the day to maintain an accurate black level with valve characteristics drifting around. Not to mention leaky paper capacitors and noisy carbon composition resistors. Just keeping the whole show running would have been difficult enough.

If the TV had simple AGC - many did, including RTV&H - DC coupling was a bit of a waste of time anyway.

They did publish an article with a modification to their designs to add a DC restorer. This circuit drove G1 and provided a variable amount of DC restoration.

I had always understood that colour sets must have true DC, but recently see this was, alarmingly, not always so!

Yes that is true for colour.

I don't recall ever seeing a PAL TV that didn't have DC coupling / restoration. However, some older designs would perform beam current limiting (necessary to avoid embarrassing the HV circuits and the CRT itself) with the "Brightness" control, thus defeating the black level maintenance when the beam limiting was invoked by high contrast settings. So if you mis-adjusted the TV you'd lose correct black level.

The "better" designs limited the contrast, i.e. the white level and the black level stayed constant.

As mentioned in Wayne Bretl's paper on 17 problems with early colour transmission, one of the 17 was "lack of DC restoration following first designs". I think he was referring to RCA's next designs following their first CTC-2 chassis (and of course pro studio monitors) where no expense was spared. But the following consumer designs necessitated cost cutting thus the CTC-4 and 5 chassis (1955 & 1956) seem to do the capacitor-across-high-resistor thingy.

Do that on a colour TV and the colour balance shifts with scene brightness, it's a real no - no.

Did they really implement beam current limiting that way? If so, it's no wonder colour took so long to become established in the US. It was obviously a lot more than the NTSC-related hue shifts. Horrible!!

In the early days of colour here, on installation, we used to set receivers up to match a portable illuminant - D 6500 degrees K greyscale reference lamp. And by then, studio techniques had developed so that the quality was consistently good. A well-set-up PAL receiver is a joy to behold and would quickly result in all the neighbours rushing out to buy one.

Harking back to early B&W, right now I have an HMV F3 from 1957 on the bench. I had forgotten just how good this model could be, once you fix all the niggling faults that had been allowed to build up over many years. This set actually resolves out to 5MHz, via its tuner and IF no less. And of course the black level is rock solid with huge changes in APL. Bright highlights in dark scenes punch through with incredible impact.

Now, if I could find the cause of that tiny but annoying intermittent jump in the vertical I'll be happy. Time to break out the scope and try to capture it.

Here is Bretl's actual wording on this, as well as two other issues (that affected PAL viewers as well) (points 11, 12, 13 of 17.)
In fact the latter problem is with us to this day:

• Lack of full DC restoration in all but the earliest
generation consumer receivers
• Changes in receiver phosphors over the years to increase
brightness at the expense of colourimetry [Ref02, Ref03.
Ref23]
• Early adoption of high colour temperature white points in
receivers

So it was especially the popular RCA CTC-5 (1956 & 1957) that could be construed as having only partial DC integrity.

PAL viewers were especially victims of the second one as there was an 'arms race' on picture tube brightness between Japan & USA, that's why those Japanese TVs, like Rank Arena, NEC, Hitachi... had funny colour that in no way matched European phosphor formulations.

As for the final point, let's note a peculiar spec of the Japanese TV system, NTSC-J: they specified a 9300k white point: Even today flat panel Asian TVs usually have an adjustment hidden in the advanced menu for "Colour Temperature: Cool, Med, Warm." Consumers have no idea what to do with this, other than mess it up, which they can be counted on to do

Yes, the big issue was the "red" phosphor, with was much brighter but not red at all! And it also tended to be non-linear, which was fine in NTSC countries as it covered up the effect of hue errors on fleshtones but made them much too red-orange when the correct vector was displayed!

Mitsubishi made a feature of this at one stage by correcting the red optically using a blue tinted faceplate glass.

Toshiba did a special phosphor combo for Pye which was not as bright but was more accurate. I think Hitachi did the same for Philips.

But really only Philips CRTs gave accurate colours.