mac said:
i have thought about it. it seems its a cheap thing to do. right now im considering upgrading from the standard nichicons to nichicon KG golds (3x 4700uf's per rail, so 28,200uf per supply per channel). that increases the cost of the supply quite a bit. all the little bits add up after awhile.
i also thought about using IXYS bridges as well. they seem to be the best ones out there. i figure those bypassed with small value WIMA caps into a bank of nichicon KG golds with small filter caps should be about as good as it can get for the money. and its not much more than buying a pre-built supply from hypex.
cowanrg said:
I decided to not use fast recovery bridges in my amps. Some people (including Nelson Pass) believe that they aren't well suited for high current applications. IMO the bypass caps should be used across any large value electrolytic filter cap. Easy and cheap enough to experiment with...
mac said:
Fast recovery no, soft recovery yes! I believe NP uses them as well.
Unless fast recovery refers to a schottky, which Pass seemed to recommend, but said he doesnt' use them himself due to their extra cost and soft recovery likely being good enough.
cowanrg said:
You'll have to forgive me if some of this is redundant with your current supply or plans.
First of all the level of info available on what type of rectifier sounds best is shocking.
To save you the trouble:
#1 Schottky: No recombination (recovery) time at all, can resonate with secondaries, must be snubbed.
#1. SIC Carbide.. some impressive features, crazy expensive, choices limited, hard to come by, I'd say ignore them for a few more years.
#2. Ultra Fast Soft Recovery. (FRED)
The soft recovery produces a lower noise turn off than a regular SIC rectifier. Can and should be bypassed for an even softer recovery, possibly not required with T-Networks, but cheap to try anyway.
3. Never use fast or ultra fast without the soft recovery, a faster switch will only produce more ringing, that's why they should be avoided.
The problem with Schottky's or any discrete rectifier is that they're discrete.
Here is what I'm using:
http://www.ixys.com/l363.pdf
You'll note they dont' claim to be soft recovery but pretty well all FRED rectifiers are. I've compaired them to other FRED rectifiers from IXYS which were claimed to have soft recovery behavior and they came out favorably. They work great and sure have some good ratings, good for welding 🙂 They're $22USD each at partsconnexion, maybe you can find them elsewhere cheaper.
The soft recovery is indeed the main factor which makes them favorable to use. You may certainly add extra bypassing which will only produce an even softer recovery.
Notice they're in an easy to use and space saving eco pack. A prime feature of this is that all the individual rectifiers will be very tightly matched and share the same temperature as well. That keeps them working together very nicely, which you aren't assured of with discretes, without alot of extra effort, that realistically you can't really do.
I'm not sure what you're current layout is right now. Single rectifier with center tap etc.. I think you still have the floating inputs + star point?
That originally worked well for me when the seemingly recommended way produced hum. It was a cheap work around to get some sound out but ... Let's not ignore the recommendations.
I changed mine from a single bridge rectifier to dual IXYS eco packs linked to above, fully floating the supply. The only earth connection the modules have is their signal grounds, which correctly references the modules to the signal, instead of referencing the signal to the modules ground.
The cap common point still gets the power ground from the modules, but is not connected to earth at all.
I also only used 20 000uF of T-networks for a stereo setup @ 4ohms.
I have 15 000uF per rail per channel Jensen 4 poles ready to go in next, 2 in // per rail. It's a little beyond what most claim to be the sweet spot, but my transformer is also oversized so we'll see how it all balances out. It still isnt' beyond the recommended max which would lessen the power factor. I'm really not a fan of the idea of paralleling too many caps without a good PCB design, it's bound to hit that point of diminished gains, given all the added strays, and wire area. So I think you're plan for two to three in parallel is reasonable.
I've done absolutely no added snubbering or bypassing aaaaat all. I would however bypassing the secondaries with 100nF caps which could also help bypassing some of the zener noise from the bridges but without passing it along to the caps.
Give that kind of floating setup a try though and I think you'll be stunned with the difference, I sure was, and it was obvious from the very first second I powered it up, hardly something you have to strain to notice the difference.
Regards,
Chris
hum. thats some good info.
i think i might go with the IXYS bridges, and only do one per channel. ill stick with my current transformer, which is 1.7KVA, plenty for two UCD400's. if i can do it this way, ill use the dual secondaries to go to two separate IXYS bridges, each feeding a cap bank and thus powering a pair of UCD400's. so it will be dual mono except ill just be using a single transformer.
ill bypass the bridge with small value (47pf) caps, and think about bypassing the cap bank as well. ill use the nichicon KG golds, 3x 4,700uf per rail.
i think that should turn out to be one hell of an amplifier 🙂
i think i might go with the IXYS bridges, and only do one per channel. ill stick with my current transformer, which is 1.7KVA, plenty for two UCD400's. if i can do it this way, ill use the dual secondaries to go to two separate IXYS bridges, each feeding a cap bank and thus powering a pair of UCD400's. so it will be dual mono except ill just be using a single transformer.
ill bypass the bridge with small value (47pf) caps, and think about bypassing the cap bank as well. ill use the nichicon KG golds, 3x 4,700uf per rail.
i think that should turn out to be one hell of an amplifier 🙂
cowanrg said:
Quasimono? lol.... I'm running mine off the same supply too and don't find it an issue at all. I'm also running them biphase where you invert the inputs and outputs of just one module so that each module pumps the opposing power rail, instead of having them both pump the same rail.
Just so there's no confusion there for anyone, because it would fry their module, the power still connects to the module the same way, only the inputs and outputs get inverted, on one of the pair. This way they re-enforce each other.
It's proven itself to be extremely fast and a very potent supply even with just two 10 000uf T-networks.
I kind of do think it's far better to be a little more extravagant with one very well built supply with the best parts you can get, than to opt for two supplies with cheaper parts. I think most of us are faced with that sort of compromise. I wanted the most bang for my buck and I think I'm getting it.
So yeah, cool plan, I very much look forward to hearing your next review once you've got that done. If it's close to the Rotel now.... just wait 😉
BTW, there are some bright people who dont' advocate the standard way of bypassing the bridges, so it might be a real good idea to break it in without them first and then see what kind of an affect they have later. You may find yourself looking for a new way, or just taking them off again.
Just started wiring my Jensen caps
Regards,
Chris
classd4sure said:
Quasimono? lol.... I'm running mine off the same supply too and don't find it an issue at all. I'm also running them biphase where you invert the inputs and outputs of just one module so that each module pumps the opposing power rail, instead of having them both pump the same rail.
Just so there's no confusion there for anyone, because it would fry their module, the power still connects to the module the same way, only the inputs and outputs get inverted, on one of the pair. This way they re-enforce each other.
It's proven itself to be extremely fast and a very potent supply even with just two 10 000uf T-networks.
I kind of do think it's far better to be a little more extravagant with one very well built supply with the best parts you can get, than to opt for two supplies with cheaper parts. I think most of us are faced with that sort of compromise. I wanted the most bang for my buck and I think I'm getting it.
So yeah, cool plan, I very much look forward to hearing your next review once you've got that done. If it's close to the Rotel now.... just wait 😉
BTW, there are some bright people who dont' advocate the standard way of bypassing the bridges, so it might be a real good idea to break it in without them first and then see what kind of an affect they have later. You may find yourself looking for a new way, or just taking them off again.
Just started wiring my Jensen caps
Regards,
Chris
hum. looks like ill have to do my own research on this one. one of these days ill get around to getting them back in the shop for tweaking. i should probably just order a bunch of parts and swap them in and out and see which ones sound better. but i just dont have THAT much free time and dont want to put that much energy into it (some DIY'er i am huh!).
i guess i do have some time before the cases will be finished, so i might as well mess around with it. i just wish there were definitive answers with this stuff.
cowanrg said:
I'm doing that research right now, and I can tell you that there are about a dozen opinions on each of these elements in the long UCD400 thread and another dozen in the 180 and 700 threads.
It took 3 days to browse through those threads, even though I followed the initial 40 pages when they were started a year ago. 😀
I'm still on the fence with the 400/700 decision. If it was just a cost difference, I think I'd jump on the 700s just to make sure I don't end up with amps I feel I shouldn't have built. It's really the lack of ucd 700 info out there. Getting the core parts for the 700s is actually easier in the US than for the 400s, at least if you want to maximize the output. I can't find the 42V toroids so I am already dropping a few watts from the potential of the amps that may or may not be close to what I need. What I don't have is a DC protection circuit for an amp that can put out 28A per channel... The Vellemann kit, even if manually upgraded with huge releays, doesn't quite strike me as a nice solution with releys in the output signal path, since the Hypex HG supplies seem to get by with a few caps and resistors.
The endless thread on DC protection over in the SS forum doesn't appear to deliver any conclusive PSU-side Mosfet solution or anything else without major question marks. Also, most people in that thread are talking over my head and I don't want to mess with AC power when I don't understand what I am reading 😕
What are you doing in that regard? Ignore it? My Maggies don't have fuses anymore...
So I'll sit back another week or two and keep pondering. I still have to finish a preamp project anyway.
Peter
pburke said:
duh - looked at the data sheets on the Hypex site. Now everything is clear. And, just 24 hours after the above post, I think I will be building UCD 400s.
Peter
uh-oh, you are quoting yourself!
i honestly think the 400 is the way to go for maggies. unless you are driving a large full-range speaker, the 700 is overkill.
i need to make a decision where i want to go with my project. i just got a quote back from the guy who is going to be finishing off my cases and it was HIGH. i mean, its reasonable considering the amount of work, but it aint cheap. sigh. its always a money issue...
i honestly think the 400 is the way to go for maggies. unless you are driving a large full-range speaker, the 700 is overkill.
i need to make a decision where i want to go with my project. i just got a quote back from the guy who is going to be finishing off my cases and it was HIGH. i mean, its reasonable considering the amount of work, but it aint cheap. sigh. its always a money issue...
DC output protection
I've read everything I can find on Hypex's website, and haven't come across anything showing a recommended DC protection scheme, only the statement that their power supplies have it built in. Can you point me towards whatever cleared this up for you? I'm currently running my UCD400s without any DC protection which I'd rather not do.
pburke said:
I've read everything I can find on Hypex's website, and haven't come across anything showing a recommended DC protection scheme, only the statement that their power supplies have it built in. Can you point me towards whatever cleared this up for you? I'm currently running my UCD400s without any DC protection which I'd rather not do.
In one of the posting is described how we have done the DC protection. All our Supply boards has a relay for switching off the high rail voltage. However with the UcD700 supply, we even discharge the capacitor mounted on the UcD700 board (4 pcs 220uF/100V in parallel!). And we have in the UcD700 removed the DC detecting circuit from the power supply board to the amplifier. This is more logical... 😉
Jan-Peter
Jan-Peter,
Does this mean that, in case of DC at output in a UCD400/Hypex PS situation:
A) The transformer is decoupled, and the caps still are connected to the amps?
or
B) Caps+Tranformer are disconnected from the amps?
Thanks. Case A could still kill a speaker!
PS. A question about UCD400 overcurrent protection - it is at 20A, formerly it was at 15A. Does that mean the UCD can now push >400W even into 2 ohm loads? 😱
I used this calculation:
min ((U/root(2))^2/R , R*(I/root(2))^2)
Where U is max rail voltage, I is max current, and R is speaker load (impendance)
Or am I totally wrong? 🙂
Does this mean that, in case of DC at output in a UCD400/Hypex PS situation:
A) The transformer is decoupled, and the caps still are connected to the amps?
or
B) Caps+Tranformer are disconnected from the amps?
Thanks. Case A could still kill a speaker!
PS. A question about UCD400 overcurrent protection - it is at 20A, formerly it was at 15A. Does that mean the UCD can now push >400W even into 2 ohm loads? 😱
I used this calculation:
min ((U/root(2))^2/R , R*(I/root(2))^2)
Where U is max rail voltage, I is max current, and R is speaker load (impendance)
Or am I totally wrong? 🙂
Re: DC output protection
look at the data sheets available for the 400 HG supply right off the products page, then at a few photos of those supplies. Pretty straight forward. You can always buy one and make sure you got it right. I am just looking at the schematics to make sure I can use Blackgate FKs in the supplies and not spend extra cash on slit foil caps I won't use, although I have yet to reach another level of madness to want to spend $1120 on 8 capacitors ;-). If I don't go with the BGs I may just buy the HG supplies - very clean and compact solution.
Peter
GregD said:
look at the data sheets available for the 400 HG supply right off the products page, then at a few photos of those supplies. Pretty straight forward. You can always buy one and make sure you got it right. I am just looking at the schematics to make sure I can use Blackgate FKs in the supplies and not spend extra cash on slit foil caps I won't use, although I have yet to reach another level of madness to want to spend $1120 on 8 capacitors ;-). If I don't go with the BGs I may just buy the HG supplies - very clean and compact solution.
Peter
Yves Smolders said:
based on the schematics, the relays are between the caps and the modules.
Sure would like to have more power into 2 ohm but 2x400W on each speaker will be plenty I think, even with some impedance dips to about 3 ohms in my speakers. I have to keep reminding me that I am doing this with about 400W full range right now and it doesn't clip.
Peter
cowanrg said:
Was lonely here - so I had to keep the thread going alone 🙂
Cases - I found a place in Italy that sells really nice looking enclosures of all sizes and in several styles. I think I'll be getting a full size amp case there for each side of my system. Maybe an option for you, but I guess you already have half of the enclosures paid for...
http://www.audiokit.it/ITAENG/Cabinet/HI-FI2000/HI-FI2000.htm
Peter
Re: Re: DC output protection
Jan-Peter's comments about specific component values.
http://www.diyaudio.com/forums/showthread.php?postid=754106#post754106
Bruno's comments about how the relays work.
http://www.diyaudio.com/forums/showthread.php?postid=482796#post482796
Jan-Peter's note about what relay they liked.
http://www.diyaudio.com/forums/showthread.php?postid=482986#post482986
I just don't know how to connect the pieces together, and what the spec should be on the relays. I'm actually planning on buying a pair of the UCD700s and the matching UCD power supplies, but I'd like to use my UCD400s on another system. I suppose I can just wait until I get the UCD700s; it will hopefully be apparent by looking at the traces on the PCB board and the installed components. If I can't figure it out from that, I suppose I could buy one of the HG supplies to figure it out.
Schematics, what schematics? 😕 I haven't seen any schematics of Hypex's DC protection scheme, nor would I expect Jan-Peter to post them. 😉pburke said:
Jan-Peter's comments about specific component values.
http://www.diyaudio.com/forums/showthread.php?postid=754106#post754106
Bruno's comments about how the relays work.
http://www.diyaudio.com/forums/showthread.php?postid=482796#post482796
Jan-Peter's note about what relay they liked.
http://www.diyaudio.com/forums/showthread.php?postid=482986#post482986
I just don't know how to connect the pieces together, and what the spec should be on the relays. I'm actually planning on buying a pair of the UCD700s and the matching UCD power supplies, but I'd like to use my UCD400s on another system. I suppose I can just wait until I get the UCD700s; it will hopefully be apparent by looking at the traces on the PCB board and the installed components. If I can't figure it out from that, I suppose I could buy one of the HG supplies to figure it out.
Jan-Peter said:
Your 700 PS datasheet mentions an "audiophile grade" aux supply.... 🙂 If I'm looking at the right spot it looks rather simple, how audiophile is it? Something like a 7812/7912 with a few caps or more of a jung super reg?
Regards,
Chris
- Status
- Not open for further replies.