Meridian Unplugged - Powered by MrTechGuy.co.uk Forums Discussion Meridian - Systems Power supply upgrade

I opened up my G68 and did some measuring earlier. The front space of the unit is occupied by the existing PS and an 80mm fan.

The distance from the side of the unit to the fan mouting is approx 245mm
The distance from the central board retaining struts to the back of the unit is approx 120mm
(difficult to measure accurately as there are variable sized components protruding from the front circuit boards)
The max available height is 110mm

So, not a lot of room overall without moving the fan. I would add a picture if I could find the file management button.

Cliff

PS my G68 is one of the first generation models (AXV). I don't know if there any internal mods to the later G68 models.

I posted photos of my G68D PSU some times ago: you should be able to find them with my post #148135 - 08/28/11 12:23 PM

Thanks for the measurements Cliff.

I used to have a G68, but I don't remember the insides very well. Does yours look like this Meridian G68 Internal View?

In that picture, the 245mm is presumably the horizontal distance from the right side of the fan housing to the right hand side of the unit, on the right? In other words, the space that is half ocupied by the SMPS? So if the fan could be relocated to the left, there would be about 300mm available? I ask this because the form factor for the popular HBAA-40W LPS is 260x102x63. It sounds like that would have enough space if it could encroach on the fan aperture.

And when you say "the central board retaining struts", do you mean the plastic brackets that aren't visible in the above picture, but can be seen in Meridian G61 Internal view of the G61?

So that there's 120mm available without having to cut the struts in the G68? That being the case, there would be nearly enough space for the 60W Acopian 51212T9A at 235x87x130 (which I've just bought). I have high hopes for a G68 with a linear supply...

Thanks, Nick

To those who were talking about different configurations of power supply, from previous experience I was trying not to be too ambitious on my first attempt, and just wanted to get something working.

An external supply may be the only option for the slim G series units, or any of the disc players, where half the available space appears to be given over to the disc drive. That wasn't going to be the configuration of my first attempt, though, as its difficult to justify the extra time to something that I couldn't be sure was going to work.

An external supply is also a bit more complex. I guess I'd need multi-pin XLR conenctors for the power supply connections, BUT I'd like to keep the separate return lines for the 5v & +/-12v supplies, plus a separate ground, plus the sensor lines to compensate for the additional wiring and connectors. That makes for a complicated connector.

With an external supply, I could presumably use a single supply to drive multiple units - a player, a controller and an HD621, for example. I'd be tempted to use a single large transformer feeding individual rectification & regulation modules for each equipment. It would be difficult to swap units around, then, though.

Finally, I didn't change the orientation of the supply in my V2 as I have to strip it every time I remove the PSU, plus I fitted the R3 cover to the chassis, which closes the box (bar some ventilation slots). That will have to do.

Nick

No, problem


Yes, the same or very similar




Yep



Similar, but in the G68 they are not so extended, i.e. while they taper so that they are wider at the bottom but the angle is much steeper.


Yes, the 120mm is from the bottom of the struts. With the 51212T9A there would be some scope for raising it slightly and/or cutting into the struts. It would be a tight fit and avoiding the connectors on the front boards may be the biggest problem. Adequate cooling may also be a problem.

Cliff

I've ignored this thread, but haveing just read it -realised how interesting it is.

GREAT work Nick.

Those struts look like to support PCB insertion/extraction ??
Maybe they could be replaced with metal "L" brackets, so freeing up valuable space.

I have just searched for the G68 PSU you purchased Nick, couldn't find that many suppliers.

Again, well done!!

I've just taken the lid of my G68 again and measured the distance between the bottom of the struts and the front of the case more accurately. Taking into account the protruding connectors which connect the front two boards (these are at a height of approx 85 mm above the base), my original estimate was bit generous and it's more like 117mm.

At their base, the supporting stuts are 38mm wide so thers's scope for cutting or replacing these. I wouldn't want to go within the 20mm of the circuit board so there's approx 135mm of space to play with.

Thanks for your input, Cliff.

I�d also like to get some second opinions on how to implement linear power supplies.

Firstly, although there�s a wide range of makes & models, few distributors hold much stock, and those that do are quite expensive. Used equipment is cheaper and more readily available, but although it�s proven technology, it may be rather old. Realistically, few of us are likely to modify brand new equipment, so how important is a new PSU? For anyone who�s considering fitting an LPS � would you want the assurance of a new unit, or would you be happy with a used one?

Secondly, when I fitted the LPS to my 861v2, I soldered the mains leads to the PSU, which is easy enough to access from above or below. However, the power leads have to pass through a cut-out in the backplane, and extending the cut-out to the bottom would make access much easier. There�s an earth plane that extends to within a couple of mm of the edge. I�d like to slightly shorten the backplane, leaving the ground plane at the edge. Does anyone think this would cause a problem?

Finally, linear supplies run warmer than switching supplies, and will dissipate heat at near operating level even on standby. Is it important that that equipment should be safe and reliable when left on standby continuously, or is it acceptable that modified equipment should be switched off overnight?

Thanks, Nick

Oops. I was just putting my v4/LPS together, and I realised that the comparison I made between the stock v4 and modified v2 wasn't a fair one. Everything else hadn't been equal after all.

The v2 had an OA02 analogue card, while the v4 had an OA03 card. This obviously tipped things in the v4's favour. Please accept my apologies for the misleading result of the shoot-out.

Nick

OK, I finally got the v4 working with a linear power supply as well. Instead of the R3 Technology 00-3516 supply, I used a Lambda HTD-1. On the face of it, it was identical, but there was a crucial difference. The Lambda supply was almost silent. I could hear the v2 from across the room, but I had to put my ear up against the v4 to hear anything. What a relief. I have an Acopian and a Condor supply on their way as well, so I still have a bit more playing to do. Perhaps my R3 PSU wasn't as good as it ought to have been, but it was new. The next bit of advice is obviously not to get an R3 Techology power supply. That's a shame, as it's British, but maybe that's what you get when you buy the cheapest (and lightest!) product.

Audio-wise, the LPS made the same sort of improvement with the v4 as it did with the v2. The transparency and clarity were transformed, and the v4 performance was restored to lead the v2 again. It seemed to have a bigger envelope - bigger frequency and dynamic range, and larger soundstage, yet still well-resolved within that envelope. There was nothing muddled or out of control. However the v4 LPS seemed to have a new character to it - it was warmer and more refined, more rounded, solid and tangible. It was very listenable and engaging. A lot of things I'm describing are like the Meridian "house" sound; its all there, just much more of it.

Again, the quantitative gap between the v2 and v4 was much the same with LPS as it had been with SMPS. My feeling is that the v4 in isolation represents a distinct step-forwards over the v2, and so too does the OA03 card over the OA02. However, those improvements pale in comparison with the difference that the LPS makes, and now I have a quiet-running controller to boot. Very happy camper now.

Anybody want to buy a slightly-used R3 triple output linear power supply?

Nick

It had been my intention with the 861v4 to repeat some of the original power demand measurements that I did with the v2, and to see what the consumption of an LPS is. I measured min and max consumption as before, but only found a variation of about 3% between best and worst case, so all these figures are worst / typical / operating case.

With the stock v2, I had measured 33W input and 24W output from the SMPS. That means 9W is dissipated in the SMPS, which is 73% efficient.

With the LPS in the v2, I presume that the load was still 24W (I didn't measure it) but now the consumption from the mains was 360mA x 240V = 86W. That means the dissipation has gone up to 62W, and the efficiency is down to 28%. LPS manufacturers claim 45-55% efficiency, but only at full power (this one is running at third power).

The LPS in the v4 was interesting. I didn't measure the (internal) load currents, but the mains load was 230mA x 240V = 55W. I doubt the Lambda supply is any more efficient than the R3 Technology, and both of them are presumably less efficient with less load, so that leaves two unknowns. I don't know the internal load for the v4, and I don't know either the dissipation or the efficiency.

Either way, the electrical load of v4 must be significantly less than the v2. If the efficiency is the same, the load will have dropped from 24W to 15W. The efficiency is more likely to be lower, so the actual load will be less than that - maybe even 12W? Interesting stuff, and it suggests that the newer equipment draws less power than older, in spite of the increased DSP power, and the PSU demands may not be as high for everything as I put in my chart.

Using a separate transfomer to feed individual rectification/regulation boards is a longer term option, so I measured the secondary outputs of the Lambda supply. The 5V board took +/-9.3V, and the 12V board took +/-18.9V. The measurements of the R3 Tech supply in the V2 were much the same: +/-10V for the 5V board, and +/-19V for the 12V board. This points to both supplies having similar design and efficiency characteristics, and the transformers were a very similar size. (These are AC RMS measurements of course.)

That's just for the record, in case I get really carried away and stick a 1kVA toroid in a separate box, or try and fit an LPS into an HD621. One day, maybe. I'm very curious to hear how the Ruby 808.3 sounds. Based on what I heard, it should be a significant step fowards.

Nick

Edit: There are quite a few easy pieces to the jigsaw, and I've tried to place them when where I can. Unfortunately, I forgot to measure the DC currents on the output of the SMPS on the V4, and now they're all soldered up, I'm reluctant to do that. However, the DC outputs on the V2 use spade connectors, and the +5V/+12V/-12V loads drawn from the LPS are much the same as for original measurements on the SMPS, that I posted in the big spreadsheet.

However, more interesting than that was what happens when I disconnect all the DC outputs. All power supplies are most efficient at full load, and zero percent efficient at zero load, but what about the actual numbers? With zero output, the LPS continued to draw 220mA @ 240V, or 55W. This is a 70W power supply that is around 50% efficient at full load. Therefore we have a few points of reference for this PSU:

In general:
Consumption = Load + Dissipation
At zero load: 55W = 0W + 55W
At mid load: 86W = 24W + 62W
At full load: 140W = 70W + 70W

So the power dissipation doesn't vary very much - less than the load does. That suggests that having an oversize linear power supply will probably lead to greater power dissipation at whatever the equipment's actual load level is. And a small, harder-working, supply will dissipate less heat than a big supply that's cruising. Looking at the power measurements in the Wiki, I noticed that many of the other equipment has total consumption that was similar to the 861V4, and hence less than the 861V2, which drew just 24W internally. That suggests that some of the equipment that I thought might cause problems accomodating a linear PSU, like disc players, or the G-series units, may not need big power supply after all.

There�s a ballast resistor that loads the capacitor board on the 861, 808 & 800 power supplies, and it�s been bugging me. I assumed it was there to keep the SMPS happy, but I�d like to remove it with the linear supplies (and I wanted to make sure there wasn�t a good reason why I shouldn�t).

It�s a small 33 ohm power resistor that�s bonded to the base of the chassis, and connected to the capacitor board by two short, green wires. It�s not visible in many of the internal pictures, but you can see it clearly in this one:

http://www.Meridian-audio.info/viewphoto.php?compid=50&photoid=2600

The resistor is pretty much in the centre of the picture, between the PSU, fan and capacitor board. It sits on one of the 12V lines, so it pulls about 360mA and dissipates over 4W. With a linear regulated supply, the total dissipation is higher than that due to the inefficiency. My estimates of actual power consumption (of the newer units, at least) are going down, so this load constitutes a significant fraction. My V4 consumes no more than 15W internally, so removing the resistor would make it 11W, tops.

Linear supplies work OK with no load � I tested mine the other day when I was measuring the no-load consumption. However, SMPS often need a minimum load just to keep them ticking over, and the start-up sequencing may be a critical factor here. The +5V line has no ballast resistor, so I presume that comes on line first, and the delayed load from the 12V line might cause the SMPS some problems. If you look at page 2 of this XP Power datasheet you can see that multi-output models like the ECM40UT31 need to have a minimum of 100 � 500mA load on one or two of the lines.

http://datasheet.octopart.com/ECM40UT31-XP-Power-datasheet-7541498.pdf

I think this is a particular requirement for SMPS, so I think it�s safe to leave the resistor disconnected when fitting an LPS. That will leave the equipment running cooler, and it also means the real power consumption of the newer equipment (ie: not the early 800�s & 861�s) may be quite low. This gives more flexibility with which power supplies can be used.

Nick