Hi, I bought an RKB-D8100 on eBay - sold as an RKB-8100 but it does have the digital inputs.
It runs with both fans on all the time, even after upgrading to the latest firmware, and the temperature levels off at 38-40C on both heatsinks and also both power supplies, even with the top off and no music playing. The heatsinks certainly don't feel that hot to touch, so maybe the thermistors are bad.
Anyway I thought I'd take it apart and see if its salvageable.
The 8-off amp modules are UcD250LP-OEM, as documented here: https://www.hypex.nl/products/amplifier-families/ucd-family/ucd250lp-oem. Each side of the amp has a motherboard and a tubular heatsink with a fan at the back.
There are two power supplies marked 'A400' and I believe them to be SMPS400A400, given the documented voltage for the '250s.
There is also a Rotel-branded power controller that provides 3.3v (I believe, per tiny writing) to the control CPU board, and has signal back from it that I believe controls a relay that shuts off the SMPS modules in standby.
The input sockets are all connected directly to a board that has a +-12V power connection from one of the motherboards.
The control CPU board is under the input board next to the RS232 socket. I haven't got a good look at it yet, but it has connections to the master power controller, the amp motherboards, the trim pots on the front panel, and the input board - presumably to manage the selection of digital input vs analog input and I guess also the rather pathetic volume control from software and the trim pots. Anything under 50 is too quiet to hear and the volume ramps up between settings 80 and 96.
Anyway, my plan is to remove the motherboards and the control CPU board.
I might remove the power supply control/distribution board.
I will try to retain the input board if I can figure out how to lock to analogue and the max volume. Otherwise I'll try to remove it and re-use the back panel with new RCAs.
I'll get a step-down DC/DC converter to run the input board and whatever else I have at line level.
I'll get some DRV134 balanced driver boards.
I might try to get some digitally controlled line level volume controls.
And I think get 6off retail UcD400HxR. I don't need all 8 channels. I don't actually need mich power for mids and tweeters but I think the '180s will not be happy with those PSUs, and they are only marginally cheaper.
I'll probably wire up an Arduino or similar to RS232 and some temp sensors and the power distributor (if retained) and the volume controls (if I go for that), then I can control it all from the PC that will be running the crossover. I have plenty of RPis of assorted vintages but I think they will be overkill and actually harder to use in this case.
Has anyone tried anything like this before? See big problems I haven't thought of?
There is definitely risk in being able to reuse the input board without the Rotel management CPU. It might be simpler to ditch it completely in favour of some new buffers, if the DRV134s are not sufficient.
Also it might be that I have thermals to deal with - I really want to lose the fans. I can understand why they are there, given the unit's role as a rackmount device.
I have noticed some rather cheap Chinese UcDs, claiming '500W' (huh?), here, as both mono and stereo. Tempting to try them, although clearly without HxR: https://www.ebay.co.uk/itm/175640477619
It runs with both fans on all the time, even after upgrading to the latest firmware, and the temperature levels off at 38-40C on both heatsinks and also both power supplies, even with the top off and no music playing. The heatsinks certainly don't feel that hot to touch, so maybe the thermistors are bad.
Anyway I thought I'd take it apart and see if its salvageable.
The 8-off amp modules are UcD250LP-OEM, as documented here: https://www.hypex.nl/products/amplifier-families/ucd-family/ucd250lp-oem. Each side of the amp has a motherboard and a tubular heatsink with a fan at the back.
There are two power supplies marked 'A400' and I believe them to be SMPS400A400, given the documented voltage for the '250s.
There is also a Rotel-branded power controller that provides 3.3v (I believe, per tiny writing) to the control CPU board, and has signal back from it that I believe controls a relay that shuts off the SMPS modules in standby.
The input sockets are all connected directly to a board that has a +-12V power connection from one of the motherboards.
The control CPU board is under the input board next to the RS232 socket. I haven't got a good look at it yet, but it has connections to the master power controller, the amp motherboards, the trim pots on the front panel, and the input board - presumably to manage the selection of digital input vs analog input and I guess also the rather pathetic volume control from software and the trim pots. Anything under 50 is too quiet to hear and the volume ramps up between settings 80 and 96.
Anyway, my plan is to remove the motherboards and the control CPU board.
I might remove the power supply control/distribution board.
I will try to retain the input board if I can figure out how to lock to analogue and the max volume. Otherwise I'll try to remove it and re-use the back panel with new RCAs.
I'll get a step-down DC/DC converter to run the input board and whatever else I have at line level.
I'll get some DRV134 balanced driver boards.
I might try to get some digitally controlled line level volume controls.
And I think get 6off retail UcD400HxR. I don't need all 8 channels. I don't actually need mich power for mids and tweeters but I think the '180s will not be happy with those PSUs, and they are only marginally cheaper.
I'll probably wire up an Arduino or similar to RS232 and some temp sensors and the power distributor (if retained) and the volume controls (if I go for that), then I can control it all from the PC that will be running the crossover. I have plenty of RPis of assorted vintages but I think they will be overkill and actually harder to use in this case.
Has anyone tried anything like this before? See big problems I haven't thought of?
There is definitely risk in being able to reuse the input board without the Rotel management CPU. It might be simpler to ditch it completely in favour of some new buffers, if the DRV134s are not sufficient.
Also it might be that I have thermals to deal with - I really want to lose the fans. I can understand why they are there, given the unit's role as a rackmount device.
I have noticed some rather cheap Chinese UcDs, claiming '500W' (huh?), here, as both mono and stereo. Tempting to try them, although clearly without HxR: https://www.ebay.co.uk/itm/175640477619
For what its worth - I found that the device will run with the fans disconnected. But the temperature on the heatsinks attached to the amp modules rose to 48C and 44C with the top off and the room at 22C, and no music or attached loudspeakers. More surprisingly I turned 7 off the 8 amps off through RS232 and the temperatures did not fall.
It seems incredibly high power draw. So I think its days are numbered. Hopefully the power supplies are OK if repurposed.
It seems incredibly high power draw. So I think its days are numbered. Hopefully the power supplies are OK if repurposed.
What IS the power draw? Doesn't take many watts to heat something up a little bit. If the amp is otherwise working OK then it's a feature that you don't like, not a problem as such. Could do the traditional computer case route and see about heat pipes and larger fans or water cooling but that's a whole different can of worms.
Hi Jerms. The UcD250LP documentation suggests that the idle losses are typically 4.6W with external VDR and 6.6W with internal VDR. So its not unresonable for the 8 modules to be wasting over 50W, and I guess pushing 60 total given the SMPSs. I can see your point, but it seems wrong to me that the temperatures would rise so much in what was effectively open frame.
Perhaps an approach that might help would be to start by removing the Rotel control circuitry and retaining the tubular heatsinks and UcD250LPs but replacing the control circuitry with something that can control Nocutua (or similar) PWN 60mm fans. I can control remove two of the UcD modules - I only need 6 channels.
That might be more cost effective, if lower performance, that gutting the system except for the power supplies and using the retail modules. I could always go for those later.
Perhaps an approach that might help would be to start by removing the Rotel control circuitry and retaining the tubular heatsinks and UcD250LPs but replacing the control circuitry with something that can control Nocutua (or similar) PWN 60mm fans. I can control remove two of the UcD modules - I only need 6 channels.
That might be more cost effective, if lower performance, that gutting the system except for the power supplies and using the retail modules. I could always go for those later.
My setup uses 4xTPA3118s driven through a pair of Asus laptop bricks. Each side (2xTPA3118) pulls 2.2W almost constantly when switched on. Even if I play music through it.
I used a mains power monitor to get the numbers.
https://www.amazon.co.uk/Power-Monitors/b?ie=UTF8&node=1938287031
A bigger and better or even newer fan might make a lot of difference. As might some isolator feet and/or some form of "fan sound trap".
I used a mains power monitor to get the numbers.
https://www.amazon.co.uk/Power-Monitors/b?ie=UTF8&node=1938287031
A bigger and better or even newer fan might make a lot of difference. As might some isolator feet and/or some form of "fan sound trap".
I've taken the control circuitry off. It seems the original CPU was actually an ARMx, which seems overkill, particularly considering the somewhat rudimentart software. There is a 3.3V supply when in standby mode, and +=12V as well as 12V 2-pin fan connectors when the SMPSs are enabled.
I plan to get some 60mm 4-pin PWM fans and will probably use an Arduino - I have a step-up regulator to recover 5V and the RS232 chip is happy with 5V power and logic, though the connector ot its carrier board is labelled 3.3V.
I'll see how things go with the new fan control and existing amp motherboards with the old inputs removed. I might need some buffers, which would have been implied in the old interface board that had the DACs and volume management. And I suspect channel on/off did not power down individual UcD modules - there are no control lines from the CPU board to the amp boards that would achieve that.
I plan to get some 60mm 4-pin PWM fans and will probably use an Arduino - I have a step-up regulator to recover 5V and the RS232 chip is happy with 5V power and logic, though the connector ot its carrier board is labelled 3.3V.
I'll see how things go with the new fan control and existing amp motherboards with the old inputs removed. I might need some buffers, which would have been implied in the old interface board that had the DACs and volume management. And I suspect channel on/off did not power down individual UcD modules - there are no control lines from the CPU board to the amp boards that would achieve that.
Hi jmansion,
I'm also looking at acquiring a S/H Rotel RKB-8100 (so no digital i/ps) to use for home theatre duties, and came across this interesting thread whilst researching the amplifier.
I'm wondering if you know how easy it would be to bypass the Rotel input circuitry and feed directly into the balanced inputs of the Hypex modules?
I need to do this as my AV processor has +18dBu balanced outputs, so some 6.16V RMS - pretty much a perfect match!
If I could do this by running flying balanced cables terminated with XLR connectors outside the case then I think this would be perfect for me.
It would also open up the possibility of bridging the hypex modules, although I guess uprating the SMPSs to 1200A400s would also provide a significant power gain.
How did you get on with improving the fan control on your amplifier, I think I read in the Hypex module spec that providing an external VDR source significantly reduces the idle power consumption, though I don't fully understand that aspect yet?
I'm also looking at acquiring a S/H Rotel RKB-8100 (so no digital i/ps) to use for home theatre duties, and came across this interesting thread whilst researching the amplifier.
I'm wondering if you know how easy it would be to bypass the Rotel input circuitry and feed directly into the balanced inputs of the Hypex modules?
I need to do this as my AV processor has +18dBu balanced outputs, so some 6.16V RMS - pretty much a perfect match!
If I could do this by running flying balanced cables terminated with XLR connectors outside the case then I think this would be perfect for me.
It would also open up the possibility of bridging the hypex modules, although I guess uprating the SMPSs to 1200A400s would also provide a significant power gain.
How did you get on with improving the fan control on your amplifier, I think I read in the Hypex module spec that providing an external VDR source significantly reduces the idle power consumption, though I don't fully understand that aspect yet?
Hi @TKDdude - I suggest that you pass on by unless its very very cheap. While the PSUs are stock Hypex, the OEM250 modules are mounted on custom Rotel circuit boards which do power and signal distribution and fan control. You'd end up discarding all the Rotel boards I think - and there's not as much space inside as you'd want to add fan control. I've discarded the standby PSU facility to regain space.
Hi @jmansion, thanks for the reply - the RKB-8100 seems to go for £300-£350 on a well known auction site, which I think is a pretty good price for a built amplifier with 8 UcD250LP-OEM modules and a pair of SMPS400A400 PSUs.
My initial thoughts were to carefully cut through the legs of the pins for the +ve & -ve inputs to each hypex module, solder on the cables, and run them out the back of the amplifier from there, so ostensibly not a lot of work and minimal disturbance to the existing Rotel control circuitry. However, it looks like there are 4 pairs of cables running from the input board to the the amp motherboards, so if these are carrying the input signals it should be even simpler.
If that's going to cause a lot of other problems with the control circuitry etc then the whole project becomes a bit of a headache as you say. Not sure why I'd need to discard all the Rotel boards though if all I'm doing is bypassing the input circuitry in this way.
If the worst comes to the worst and none of this proves viable then I'll still have 10 hypex modules (incl. the PSUs) I could mount in a suitable case (something like this), and make a proper back panel with XLR i/ps. I could even add a couple more PSUs to increase the power output, although its not necessary for what I have in mind at the moment.
Are the fans a real problem - it would be lightly loaded in a domestic environment so server type fans are not acceptable.
And btw - how did you upgrade the firmware - V2.43 seems to add the optimized temperature threshold level for internal fan modes.
My initial thoughts were to carefully cut through the legs of the pins for the +ve & -ve inputs to each hypex module, solder on the cables, and run them out the back of the amplifier from there, so ostensibly not a lot of work and minimal disturbance to the existing Rotel control circuitry. However, it looks like there are 4 pairs of cables running from the input board to the the amp motherboards, so if these are carrying the input signals it should be even simpler.
If that's going to cause a lot of other problems with the control circuitry etc then the whole project becomes a bit of a headache as you say. Not sure why I'd need to discard all the Rotel boards though if all I'm doing is bypassing the input circuitry in this way.
If the worst comes to the worst and none of this proves viable then I'll still have 10 hypex modules (incl. the PSUs) I could mount in a suitable case (something like this), and make a proper back panel with XLR i/ps. I could even add a couple more PSUs to increase the power output, although its not necessary for what I have in mind at the moment.
Are the fans a real problem - it would be lightly loaded in a domestic environment so server type fans are not acceptable.
And btw - how did you upgrade the firmware - V2.43 seems to add the optimized temperature threshold level for internal fan modes.
I did upgrade the firmware, yes, but it made no difference whatsoever. Your reasoning about the raw module cost mirrors mine, although I paid slightly more than that. The're not a lot of space if you keep the stock heatsinks. I'm not in a position to comment on your plan really, I think physical aspects might be an issue. My intent is to reork with new single eneded modules, and I've dropped the standby PSU and input board completely. Its not up and running yet, but I have enough space to mount a buck converter for arduino and a couple of decent fans. Its a background project for me.
![IMG_20240102_124939258[1].jpg](/community/data/attachments/1162/1162311-2dbdbbc4b0bfd3f788a0fe973c75b0cd.jpg?hash=Lb27xLC_0_)
![IMG_20240102_125001680[1].jpg](/community/data/attachments/1162/1162313-a71829615e71549603e64a98db91b80c.jpg?hash=pxgpYV5xVJ)
The pictures are really useful, thanks. Looking at them quickly, I think it may be easier to intercept/redirect the amps signal inputs on the amp motherboard (if one side isnt immediately grounded!) - I'd prefer not to mess with the hypex modules if I can avoid it. I guess I'll just have to locate their pcb tracks and experiment a bit. And then there's the fan issue as well. Anyway - you've been very helpful and I'll let you know how I get on. It could take a while though - like you I have other priorities for the next few weeks or so.