Saturday, August 03, 2019

Calibrating the Calibrator - Colour management with Eizo CG319X

I've spent a lot of time demo'ing these monitors recently and the thing that tickles customers the most is correlating the internal sensor to an external probe.


Saturday, April 20, 2019

Electronics, disasterous career choices and good techy YouTubers...

The last few months have been a bit of an education for me; I changed jobs twice and kinda wished I hadn't - neither has been ideal. However, I wound up with six weeks with not much to do (I did a few days of freelance colour work) but I took the opportunity to finish off my home fixing bench;

After this I resorted to my usual cathartic activity of building a hand-held games console based around a RasPi and fitted into the dead carcass of an old Gameboy/Gamegear etc. I didn't make any videos of this project (unlike the Gamegear from last year;

BUT, back to the most recent build; I was really pleased how it worked out, probably the most tidy one I've done to date. I also implemented a proper shut-down script so the power button doesn't just crash the Pi (and I monitor the LiPo's voltage to avoid the system crashing that way too).

The other thing I took to whilst off was to start learning the Python programming language - if you're looking to get back up to speed with coding Python seems to cover a lot of the principles of modern scripting languages - the thing that has blown me away the most so far is the list data type - a list can have different data types within it! Perhaps all modern languages have this, but my degree in Maths and Programming in the mid-eighties did not prepare me for such things!
Can I recommend Jamie Chan's book "Learn Python in one day and learn it well" - as someone who still has a bit of C and VB experience it has bee good for me.

I have also been catching up on my favourite electronics YouTubers;
Louis Rossman - Apple repair guy and the best example of surface-mount rework and diagnostics you'll see.
Big Clive - more focussed on big volts and tearing down badly made power electronics!
Great Scott! My kind of small project builds with something of a focus on Arduino
A brother broadcast engineer who fixes stuff
The Post Apocalyptic Inventor - motors and generators are the emphasis of his builds
Dave Jones of the eevBlog is a design engineer who seems to know everything about electronics; really engaging style too.

So, not sure how 2019 is going to play out - one thing I have realised is that my plans of giving up all this broadcast engineering nonsense are perhaps closer than I thought; maybe the teaching PGCE is even closer than I thought...

Saturday, January 12, 2019

Is there an HDR mastering display that costs less than £20k? Not yet.

I was asked to take a look at the ASUS PA32U monitor - a 32" display that (when I first saw it last year) was initially an sRGB, Adobe RGB, HDR10 high-end graphics/gaming monitor. It comes in at sub £2k and so is expensive for home or office use but almost free from a film/TV perspective. Most interetsingly it could hit >1,000Cdm-2 peak white and unlike OLED displays (typ. Sony BVM-X300) it doesn't suffer ABL as more than a small percentage of the display hits peak white.

I gave the manufacturer a few pointers - essentially if it is going to have any application in film and TV it should at least support HLG and DolbyPQ. So, just before christmas a new itteration of the monitor arrived and here is a little video.

A few things to note;

  1. Terrible "blooming" near black; if it was a CRT it would be akin to bad internal reflection within the tube.
  2. The blue primary is not as good as it should be - it can't even hit 100% of rec.709
  3. Ironically, however, it does manage 84% of rec.2020 - just shows you can't get too hung up on very saturated colours,
  4. There is some sort of noise-coring going on; you should be able to disable that in the menues.
LightSpace reports;

Sunday, December 09, 2018

Compressing spoken word audio for podcasts

What a terribly lax blogger I've been over the last few months; part of that was due to a couple fo trips to Saudi to finish building a facility in Riyadh (more about that to come; did you know that protective mains earths are not mandated there?!).
Anyway - I look after a couple of podcasts (just spoken word content) and for the longest time I've been using the Conversations Network Levelator to compress the dynamic range of each clip before I import them into Audacity and then edit/normalise them before exporting as an MP3 for upload. Imagine my horror when after switching to a new (old) laptop that I installed Mojave (OS-X 10.14) onto I could no longer use The Levelator; and it's a thing - lots of folks complaining online about it, but it's abandoware so what are you going to do?

 You upgrade the OS and some software stops working...!

 People listen to podcast in noisy environments; in cars, on the bus in earbuds and in other non-ideal listening environments. So, like talk radio the best thing is to compress the signal until there is almost no dynamic range and then normalise it to less than a dB under 0dBfs. Then it's a loud and punchy as it possibly could be and it'll be usable in bad listening situations. The Levelator was brilliant at doing this, but I thought I should at least roll my sleeves up and see how good the compressor in Audacity is, and blow me down it isn't bad!

After a bit of tinkering around with the ratio and threshold I arrived at these values as being best for audio that is peaking around -10dBfs (I'm a broadcast engineer, after all!) - it produces speech that still sounds OK, but has almost no dynamic range!

In this clip the band starts at around 2'15"

Sticking this .WAV file through the compressor set up (above) produced a very usable result which I then normalised before joining it to the rest of the segments of the podcast.

It looks like there is a lot of noise, but it's the sound of a very large room and on the speakers is sounds OK.

Wednesday, July 11, 2018

Video Compression Fundamentals; a Tech Breakfast presentation

Part of the ongoing series of Tech Breakfast presentations at Jigsaw24, Golden Square.

My notes are here and I'm available for birthdays, weddings and bar mitzvahs.

Wednesday, May 16, 2018

Why I wouldn't buy a Sony BVM-X300 in 2018 (if it was my money)

For a couple of years the Sony BVM-X300 has been the 4k/HDR monitor of choice for Soho edit and grading suites. It is an OLED monitor and can (with some limits) hit 1,000 Cd/m2 peak white in it's HDR modes (which include HLG, Dolby PQ and SLog3 camera gamma).

It was the first monitor to be widely regarded as good enough for Dolby Vision mastering (and by extension Netflix deliverables). It was around the £20k mark when it launched, but by last year the price had crept up to mid-twenties and with the v2 of the monitor (which brought a second quad-SDi input and an HDMI input) which launched a year ago it now lists at £32k; but we all know nobody ever pays list for Sony...!

In recent months I've really taken to the Eizo CG3145 which (although an IPS/LCD monitor) is broadly similar in spec to the Sony, but; bear in mind the X300 suffers the following;
  1. Noise in the blacks; when I calibrate them I have to do blacks at 5% grey to get a clean reading (and my probe goes down to 0.01 Cd/m2) – the Eizo will read cleanly at 2% grey. Watch this video (hosted on my Twitter feed) - it's an X300 around 3 Cd/m2
  2. Max. 8% peak white before the orange PSU-fault LED comes on and the display starts to dim/de-saturate in HDR modes; at the recent Jigsaw24 I showed the "OLED killer" which you can get here.
  3. Two years in and several Soho X300s are now showing burn-in (particularly where the 3840-pixel UHD and 4096 pixel-4k rasters differ) 

£10k less list price and available now (the X300 is in such short supply that you can't get one in London currently) also add to the Eizo’s advantages. Integration with LightSpaceCMS (pretty much the industry standard for colour management) is very tight whereas the X300 only talks to Sony’s very clunky colour software (no LUT management, six-point calibration only). 

The Eizo easily allows LUT upload; this SLog3 (in slot 8) was imported via LightSpace.
ColorEdge PROMINENCE CG3145 is the first HDR monitor in the world to have its HDR and SDR reproducibility evaluated and awarded by the German Broadcast Technology Institute.

The Eizo has recently been certified as both a Dolby Vision mastering display as well as having Netflix's blessing.

One objection I've heard is that the X300 covers 85% of rec.2020 whereas the CG3145 only covers ~83%. It's the kind of objection that someone with a poor grasp of colourimetry makes. My answer to that is "MacAdam ellipses" - look at the Wikipedia article and tell me anyone can see the (Just Noticable!) differences.

At the recent HDR Summit at Dock10 in Media City, Salford we had more than ten HDR-capable displays.

Wednesday, March 14, 2018

Electrical Safety in Film & TV + 18th Edition; Tech Breakfast presentation

BS 7671 is the national standard to which all electrical installations should conform.  The 18th Edition IET Wiring Regulations contains important new information for all electrical installers and engineers.
In the next Tech Breakfast I focus on the 18th edition of the The Institution of Engineering and Technology (IET) regs, which will debut this summer & I review best practice for designing and constructing power distribution in machine rooms and remote data centres for film and TV.

Electrical safety is one thing no broadcast engineer can ignore. 

Tuesday, February 20, 2018

12G cabling - test results and a video presentation

I presented a recent Tech Breakfast at Jigsaw24, Golden Square. Here I detail the tests we've done across four cable types and how they perform at twelve gigabits/sec (as per SMPTE 2082-1).
We've recently taken on Leader as a manufacturer of test sets and they excel in several areas - namely UHD/4K/HDR and 12G physical layer measurements. 

 all the specs for SMPTE 2082-1

I got through all the details in the video (below and on YouTube) but you can snag my results here - if you go into the 12G folder you can see the screen grabs for all the eye patterns - the filename number related to the test line in the PDF.  The Powerpoint presentation is in there too; but if you watch the video I cut all the slides full-screen as appropriate.

 The cable types are;

SD05 - Belden 1855; otherwise known as "Image 360"
SD10 - Belden 1694
SD50 - Belden 1505; otherwise known as "Image 1000"
SD73 - Belden 7731 - about the most ungainly cable you can crimp a BNC onto!

Wednesday, December 06, 2017

Fibre for breakfast - keeps you regular...

I've have been a terrible blogger over the last couple of months; just super-busy at work, but I have done a couple of presentations in Jigsaw24's ongoing series of Tech Breakfasts.

All the slides are here.

Friday, September 22, 2017

A few notes on DolbyPQ & the new 4k AppleTV and TV High Dynamic Range.

STOP PRESS! 27th Sept. 2017 Update;

So it turns out that the new AppleTV does support HLG.

- Why did it take somebody hacking around with a firmware update to discover it; perhaps big corporations (Apple and Dolby) talk and would rather not highlight the fact?

HDR is half my life at the moment; the distinction between "Display Referred" and "Scene Referred" video is lost on most people, but is pretty central to understanding why the BBC/NHK "Hybrid Log Gamma" system is ten times more appropriate for television (non-theatrical video) vs anything based on the SMPTE 2084 (AKA Dolby/HDR10 etc) curve.

For my presentation on "intro to HDR for TV" download here.
  1. Display Referred HDR makes no sense for TV (when I say TV I mean all non theatrical video). DolbyPQ makes video dimensioned (so code values actually represent light levels) which makes a lot of sense when you have complete control over the environment you're viewing in - a theatre. To define where black and white sit (and actually assign light-levels to them) is problematic for TV workflows. Remember, you have to give the colourist/racks-engineer/domestic-viewer the liberty to set black according to the room. Although BT.1886 is commonly accepted to mean 100Cdm-2 peak white a lot of colourists drive their rooms at 80Cdm-2 and at least one film guy I know prefers to work at 60Cdm-2. Also - what happens in three years when everyone is selling TVs with specular highlights that can hit 2,000Cdm-2 and people can see the difference between PQ content mastered with peaks at 1,000Cdm-2 ( the current standard) and new content? The same will be true all the way up to Dolby's max light level of 10,000Cdm-2. Dolby at least has the benefit of dynamic metadata to allow or this, but HDR10 is static metadata and so has all the problems of display-referred HDR with none of the DolbyPQ benefits.
  2. BBC/NHK HLG is a much more pragmatic solution as it doesn't assign code-value to light-levels (when has that ever been a thing in TV?!) and allows HDR content to look good on all devices capable of displaying it; tablets, TVs, laptops etc. It also allows the broadcasters to make a gradual change to HDR. None of the broadcasters I've spoken to have any appetite for having Dolby CMUs all over the place to manage the metadata (which, being a licensed format, they would be obliged to have). HLG also tracks 1886 for most of the curve (to around 65%) which means conversion to/from is easier and even when you get it wrong the pictures look OK. It's why scene-referred video makes sense for TV.
  3. Having seen the same SLog3 (so camera HDR gamma) played out from Transkoder in both DolbyPQ (mastered at 1,000Cdm-2) next to the same machine converting to HLG with two Sony X300 monitors set for the appropriate gamma curves and the same Rec.2020 colour calibration you could not tell them apart in a blind viewing. 
  4. It's typical Dolby - they are trying to dominate the domestic space by shoe-horning their theatrical format into TVs. Broadcasters get hobbled with licensing costs, onerous upgrade requirements and pictures that are locked to whatever version of PQ/HDR10/HDR10+ they were mastered for rather than allowing the display to make the best of what it's given; scene referred pictures.
  5. The good thing about HLG is that rec.2100 ratifies it, the DVB have too. It's also trivial to upgrade HDR10-capable sets to support it (unlike PQ). I imagine it'll be the case that broadcasters will deliver HDR (for the reasons mentioned) and either you have to upgrade your TV (but pretty much all the current ones support it out of the box) OR your STB will do the conversion.
Which is why the new 4k AppleTV is a damp-squib...   

Friday, July 07, 2017

EDID will get you every time...

I was helping to configure a new board room. This one is a tiny bit different by having two nice big TVs mounted on the walls and the cable run to each is at least 30m. So - HDMI does not reliably go that far over copper cable (I know - somebody always has an example where they've made it work, but this has to be reliable!). It's probably why so many training/presentation rooms still have SVGA on the desk plate; it's reliable.
The other requirement was that with no operational changes you should be able to plug an HDMI laptop into the desk plate and " just works"!
So - Barnfind are my favourite fibre parts and so a couple of BarnMini03 & 04s with single-mode fibre between (spliced, of course!) takes the HDMI up to 80km. I used these nice wall-termination panels from ADC Krone.

I also bought one of those cheap no-name HDMI splitter/DAs from Amazon as I've had no trouble with them in the past - they even work well as HDCP removers. However in this case as soon as I plugged in the second TV the system would fall over with the "signal present" light on the DA going out - TV A or B could be fed individually but not together. Sticking the EDID analyser on the feed to the DA i/p showed the DA was generating and EDID exchange ever second or so and alternating between the two different EDIDs - first Mr. Sony then Mr. LG (different models of screens).

So, after a bit of head-scratching I put one of these on the input of the DA - it essentially answers EDID enquiries without forwarding them upwards. Works like a charm;

My next test will be to see if the DA pulling the hotplug detect pin low and triggering the erroneous EDID exchanges. For this I need to dig out one of my famous Root6 Spoof'o'matics which we had manufactured in quantity after a similar problem with Avid and how it reacts to different monitors.

Saturday, June 24, 2017

2nd RetroPi handheld build - the GamegearZero

Recently finished the second Raspberry Pi based handheld games console - this one based around a RasPi3 and a Sega Gamegear.

Wednesday, June 07, 2017

The insanity of trying to control encryption...

Politicians and social commentators keep throwing up this idea that encryption is;
  • Bad (the bolt-hole for terrorists/paedophiles)
  • Somehow controllable (back doors, side doors) by the state.
It's clear that people like Amber Rudd either have no idea (or more likely choose to have no idea) about the nature of data encryption. If you want a little primer I did an intro around a year ago and so thus armed you can explain the difference between symmetric and public-key crypto like a pro!
A few points are worth noting;

  1. Cryptography is maths - all of the algorithms currently in use are published and a decent proportion of computer-science or maths graduates could implement them in code. 
  2. WhatsApp and everyone else that offers end-to-end crypto probably rely on the underlying crypto-primitives provided by the OS - only a fool tries to re-invent the wheel (particularly WRT cryptographic functions)
  3. If you could (and most crypto experts doubt it is achievable) devise a secure public-key algorithm with a back/side-door access how can we trust any public body to not let the private back-door key get out? Five years down the line we discover that some other nation-state has had access to all the private conversations? There is much precedent for this; remember last month's NHS attack was done with code written by and subsequently lost from the NSA. Examples of large governmental bodies loosing data they really didn't want to mislay are legion.
  4. Compromised encryption & identity algorithms will spell the end of eCommerce. No bank will want to expose themselves to that kind of risk.
  5. How do you oblige software writers (who may be anywhere in the world) to use the crypto-crippled algorithm?
  6. How do you oblige "bad guys" to use the crypto-crippled software?
The outcome will be that only people who aren't concerned about security will use the crypto-crippled version of the popular chat/speech apps. Encryption exists outside of laws & countries and people who want privacy (for whatever reason) now have the means to achieve it. No nation state can now stop that.

WRT point 3 (above) I have heard non-technical people say something like "Silicon Valley is full of very clever people - they can figure it out. We were able to put a man on the moon fifty years ago..."
Well, putting a man on the Moon is one thing, putting a man on the Sun is entirely different - and that's what you're asking for, whether you choose to believe the people who actually understand cryptography or not.

Tuesday, June 06, 2017

The 18th Edition is almost upon us...

It's been nearly a decade since the 17th Edition came into force.

Currently at the committee-stage the 18th Edition will come into force at the end of 2018. The request for comments runs through until the end of August and is well worth signing up to the BSI to read the proposal and have a say if you see anything of note.

The most interesting change (for me) is section 8 - energy efficiency. All the changes to sections 4 onwards are better definitions and tighter specs for RCDs and earth leakage etc which are all very important but build on principles that are well established.

I'll post more as I get familiar with the draft, but here is the intro stolen from the IET site;

New section - energy efficiency 

The worldwide need to reduce the consumption of energy means that we have to consider how electrical installations can provide the required level of service and safety for the lowest electrical consumption. The draft proposals enable a client to specify the level of energy efficiency measures applied to an electrical installation. Installations can also be awarded points for energy efficiency performance levels, for example, transformer efficiency. These points can be added together with points for efficiency measures to give an electrical installation an efficiency class, ranging from EIEC0 to EIEC4, depending on the number of points awarded. The new section will cover several energy efficient areas, such as electric vehicles, lighting, metering, cable losses, transformer losses, power-factor correction, and harmonics.

Thursday, May 11, 2017

Sony VPL-VW320ES Projector colour profile

A recent trip to Edinburgh for my colleague at Root6 Scotland had me calibrating a new Sony SXRD-type 4k projector. Here is how it behaved from a colour profile point of view.

Monday, March 13, 2017

Calibrating monitors for HLG-1.2 High Dynamic Range working

I've now done a couple of Sony BVM-X300 (4k/HDR OLED panel) monitors for HDR-TV deliverables to the BBC. The Beeb are behind the Hybrid Log Gamma standard, currently at 1.2 for broadcast. It's benefits over Dolby's 10-bit version of their Vision HDR system (also called DolbyPQ) are many and rather splendidly the DVB as well as most manufacturers are now behind HLG. I'll write further about why a scene-referred system is a better bet for domestic TV than a display-referred one (like DolbyPQ) but I just wanted to get down some notes on monitor calibration for HLG.

So first up you need to get the monitor into HDR mode (the correct one! As of v.2 firmware the X300 supports several; the camera gammas; CLog, CLog2, SLog & SLog3 as well as the deliverable standards; SMPTE-2084 (DolbyPQ) as well as SMPTE-2100 (HLG 1.2).

You'll notice the EOTF setting (lit. "Electronic - Optical Transfer Function") is NOT set for HLG 1.2; this available but in the current v. 2.0 firmware it is wrong. Select HLG (Variable), click the know again and dial in 1.2

Next up display a 50% grey field (100% can't be done as the monitor power-limits to stop 1,000Cd/m2 being displayed across a lot of the screen) and point your probe at the display. Remember that unlike DCI-P3 we are still at the 6504K white point of tele. However - at 50% grey we should be seeing low-50's Cd/m2 being emitted from the screen. 

You might have previously heard Sony's "best practise" (sic) advice of setting 50% grey to 100Cd/m2 but this is wrong. Look at the graph; this is the result of profiling an X300 (and they all seem to do it consistently) and notice what happens over the last couple of hundred Candelas at the top end. If you set 50% grey to 100Cd/m2 you get loss of detail in the specular highlights. For the correctly 1.2 HLG curve you need to set 50% to around 50Cd/m2 to 55Cd/m2- remember the Y-axis on this graph is logarithmic. 

The one display I had previously followed the Sony advice for was sufficiently out of it's linear range that the dark-greys drifted red minutes after I finished adjusting the monitor and (like all OLEDs) the noise in their blacks is sufficient to make the final tweak to 6504k hard work - the Klein probe was having to average over 32 reads to accommodate the OLED panel. Remember - the K10A is accurate to less than 0.001 Cd/m2 and not the 1.0 Cd/m2 it reads at 10% up the curve of HLG 1.2

So - once you have those under your belt you can PLUGE the display to set blacks correctly for the room (what did I say about scene/display referred?!) and then set about getting the deep greys and the 50% colour balance correct. I've done this now on two X300s with BBC R&D engineers in attendance and they have given this method their blessing!

For a probe I was using a Klein K10A with Klein's own ChromaSurf software. If I get the chance to profile an X300 for myself I will do it with LightSpaceCMS - still the choice of champions for display profiling and LUT-building. They also have an excellent article introducing HDR.

Friday, February 24, 2017

My favorite music podcasts

For me the last ten or twelve years has seen an almost 100% move to on-demand radio listening. The only live radio I listen to is the Today Programme on Radio 4 whilst having breakfast; I do put Six Music on at the workshop for the wiring guys, but for me (even shows produced for radio) are consumed as downloads.  Here are the things about music that I never miss.

  1. BBC Mastertapes - John Wilson presents artists talking about how they recorded a classic album. The show slips in two ("The A Side" and "The B Side") with Wilson doing the interview in the first half hour and then the assembled fans asking questions in the second. Wilson is clearly a music fanatic and never fails to engage with the artist. It comes from BBC Maida Vale and there are currently 74 hours for you to catch up on! Billy Bragg's "Talking with the Tax Man about Poetry" was the first episode back in 2012 and the glorious thing about it is I've discovered some records I never knew about and been really surprised by how much I enjoyed artists I thought were not my thing. Rufus Wainwright is a prime example. The artists always do a minimal arrangement of the well known numbers off the album their talking about and so of course the format tends to favour performers who can do it live.
  2. Between the Liner Notes - A podcast (not produced for radio) that takes some aspect of the music industry and devotes an hour to it. Really engaging and has a big radio rockumentary production feel. For me - a Gen X'er coming of age in the eighties the episode on MTV is required listening.
  3. Music City Roots - this is a "proper" radio show out of Nashville. I like it because Mike Farris has a show (which sometimes makes it into the podcast feed) but the main man is Jim Lauderdale who is a real champion of Alt-country and Americana.
  4. Sodajerker on Songwriting - Sodajerker is a songwriting team from Liverpool in the United Kingdom. Founded by co-writers Simon Barber and Brian O’Connor who, like most jobbing musos, have to suppliment their work with teaching and podcasting. They have had some great guests in their (so far 97) shows. In classic podcast fashion the length varies and the best one so far (for my money) has been Paul Simon although I've enjoyed all of them (check out ep.80 - Marcus Miller).
  5. The Bored-Again Christian podcast - a notable mention even though he stopped making shows five years ago. I discovered a few artists via this podcast; Sufjan Stevens and Danielson to name a couple.
Podcasting is the answer to the dumb mass-market HeartFM/TalkFM/LBC type radio. It's no coincidence that you don't get anything like this kind of treatment of musicians and who they do on commercial radio.

Friday, February 17, 2017

My GameboyZero - build no. 1

I've been building a little RaspberryPi-in-an-old-Gameboy project for running lots of old console and handheld games. RetroPie is a superb project that unifies all of this and makes it very easy.

See my first video here.

3.5" Analogue Composite LCD
3w Class D audio amp: 
Great Scott!

Monday, January 16, 2017

LUTs are sometimes not the answer.

As I often tell people; a LUT can only reduce the dynamic range of a display. For the most part that needn't matter, particularly if you have the whole of the DCI-P3 (or a decent chunk of Rec.2020) in your monitor. Applying a look-LUT to simulate a delivery style is one thing but increasingly people see LUTs as the first answer to monitor calibration rather than getting the display to as close-as-possible before profiling/creating a LUT. Hugh and I did a podcast on the subject.

A problem I've recently discovered with a monitor's internal Rec.709 LUT is that although the monitor has a huge gamut in it's native mode (which you can see from this recording of ChromaSurf's output) but the 709 presets have trouble.

Notice how it can reach a full green value of 0.1879, 0.7317 (Rec.2020 calls for 0.170,0.797); you'd think there would be absolutely no trouble getting Rec.709 right, and in fairness the primaries are fine. BUT, when I use both primary and secondary colour ramps;

I wind up with some distinctly funny looking banding in the secondary colours.

On this 'phone photo it is particularly noticeable in the yellows, but it's there in the cyan and magenta ramps as well. The fault isn't there when the monitor is in native mode (or indeed P3, NTSC or EBU), only Rec.709 (ironically the only colour space we really need for TV!).
So, I'll have to profile it in native mode and spin a 17-point 709 LUT as the one from the manufacturer is clearly got problems.

Another way of stress-testing a LUT is to use the TrueColor LUT stress test image.

Thursday, December 22, 2016

Power supply toroid in Mackie SRM450 PA loudspeakers

A friend brought one of these very common PA powered loudspeakers over to the workshop. An internal T4A mains fuse had blown and when I replaced it and re-powered there was a very loud 50hz hum for a few seconds and the fuse in the power cord blew. I assumed it was the audio path that was making the noise and I assumed that one one the probably faults was a short/open de-coupling capacitor that was letting 50hz through to the audio stage.

I found the schematic at Mackie-SRM450 (although this is for the rev.C of the board, but the PSU appears not to have changed).

So - I made sure both fuses were good and with Dave Jones's words "thou shall always check the rails" in my ears I disconnected the output of the large toroid from the two rectifiers/smoothing caps and tried again. I figured I'd make sure the o/p of the transformer was good (there is some de-coupling on the primary). However I got the same loud 50hz for a couple of seconds and one of the fuses failed. The toroid was also noticeably warmer than it had been! So - checking the DC resistance of the primary side showed it to be less than six ohms (so it was pulling an inrush of at least 40A!).

A quick flail around the web showed that the transformer is a known weak point of this design.

 The real bummer is that the transformer isn't stocked by Mackie (or indeed anyone else) and the folks at Save My Light only do a minimum order (ten pieces) from the Chinese factory that wind them when they have enough orders (and the chap there told me he sells an average of two a year).
So, do I just wait or pay him the thousand quid to have ten made?!

Time to keep my eye on eBay for a hopefully working second hand replacement?

UPDATE 16/01/2017: 

So after lots of flailing around the web and finding a few reclaimed ones for more than £150 I came across AJAudio on eBay (Alex Mathew was super-helpful) and he sold me one for £99.
Job done.

For info - the failed toroid had around 5 ohms DC impedance across the primary whereas the replacement has 8.5 ohms.

It's been shaking the workshop with loud music all afternoon!