Hi,
I'm not OPC but I believe the Heatsink is an Ohmite WA-T247-101E. I'm currently using it for my amp's PSU rectifiers. I didn't realize till OPC did it that you can use it for the LME49830.
@Hypertune:
i would have the SMPS on the top layer, having the amps on top like that you will not be using the convection of the heatsink very well at all
i would have the SMPS on the top layer, having the amps on top like that you will not be using the convection of the heatsink very well at all
The heatsinks do not dissipate as much heat when the devices are located in the upper half of the backplate.
For best dissipation capacity, locate (a single row of) devices at ~ 40% of height up from bottom.
For best dissipation capacity, locate (a single row of) devices at ~ 40% of height up from bottom.
thanks for filling that out Andrew, i'm sure Mark would have understood, but i probably wasn't clear enough just saying to move the SMPS to the top (inferring rather than stating to move the amp section down the bottom)
Thanks Andrew & qusp.
In all honesty, the SMPS might be more trouble than it's worth. One of my goals is to make the enclosures nice and compact and it seems that heatsink area becomes the limiting factor.
I'm sure Andrew isn't the only one questioning how hard these devices can be driven but one thing's for sure and that is the cooler we can keep them the better.
Having a PS that needs additional cooling might not be the best idea in this application.
In all honesty, the SMPS might be more trouble than it's worth. One of my goals is to make the enclosures nice and compact and it seems that heatsink area becomes the limiting factor.
I'm sure Andrew isn't the only one questioning how hard these devices can be driven but one thing's for sure and that is the cooler we can keep them the better.
Having a PS that needs additional cooling might not be the best idea in this application.
That heat sink may be far to small.
OPC's initial measurements suggest a possible quiescent of 350mA per module for lowest distortion. That represents a lot of heat. To give you some idea, the Hafler DH200 used 4 devices one one heat sink, that is the same number of dies as in one OPC double die module, and, IIRC, the Hafler that I measured was also using about 350mA for one channel. The heat sink is 235mm front to back x 120mm high with 50mm thickness and at idle was maybe 50°C in a 23°C room.
Your suggesting two modules on what looks like a smaller heat sink.
Mosfets need to be kept fairly cool, eg 40°C, or their resistance goes up quite a bit and they lose their power output. I forget the numbers just now but they're something like at 75°C they are about 1/2 power. Someone will no doubt be able to post the correct figures. Their resistance, known as the r on, starts to increase at about 25°C. At loud listening the heat sinks will probably be hotter than at idle.
For 8 single die mosfets on one heat sink, equivalent to 4 double die alfets, I use, IIRC, something around a 0.25°C per watt heat sink. That's about 40°C and that's with a quiescent of only 80mA per pair (pair means one N channel and one P channel) so 320mA at idle into one heat sink.
It's also good to keep your electrolytic capacitors cool. It can possibly lengthen their life and I think I'm right in saying it's keeps their ESR down which is what you want to do for audio PSU caps as audio requires them to deliver music frequencies as low as 20 Hz, it's not just about ripple removal at 120Hz.
I may or may not also be correct in saying that you need low output impedance at 20Hz if you want good bass and no matter how frequently you recharge then, ie, from a SMPS, little caps just don't have that.
The low frequency ESR climbs hugely at low frequency, even in big caps, hence paralleling a load a big caps which always, in my experience at least, delivers better low frequencies. And good audio big caps can be in a different league to normal PSU grade ones. They can be much better sounding as well as giving much lower punchier bass.
OPC's initial measurements suggest a possible quiescent of 350mA per module for lowest distortion. That represents a lot of heat. To give you some idea, the Hafler DH200 used 4 devices one one heat sink, that is the same number of dies as in one OPC double die module, and, IIRC, the Hafler that I measured was also using about 350mA for one channel. The heat sink is 235mm front to back x 120mm high with 50mm thickness and at idle was maybe 50°C in a 23°C room.
Your suggesting two modules on what looks like a smaller heat sink.
Mosfets need to be kept fairly cool, eg 40°C, or their resistance goes up quite a bit and they lose their power output. I forget the numbers just now but they're something like at 75°C they are about 1/2 power. Someone will no doubt be able to post the correct figures. Their resistance, known as the r on, starts to increase at about 25°C. At loud listening the heat sinks will probably be hotter than at idle.
For 8 single die mosfets on one heat sink, equivalent to 4 double die alfets, I use, IIRC, something around a 0.25°C per watt heat sink. That's about 40°C and that's with a quiescent of only 80mA per pair (pair means one N channel and one P channel) so 320mA at idle into one heat sink.
It's also good to keep your electrolytic capacitors cool. It can possibly lengthen their life and I think I'm right in saying it's keeps their ESR down which is what you want to do for audio PSU caps as audio requires them to deliver music frequencies as low as 20 Hz, it's not just about ripple removal at 120Hz.
I may or may not also be correct in saying that you need low output impedance at 20Hz if you want good bass and no matter how frequently you recharge then, ie, from a SMPS, little caps just don't have that.
The low frequency ESR climbs hugely at low frequency, even in big caps, hence paralleling a load a big caps which always, in my experience at least, delivers better low frequencies. And good audio big caps can be in a different league to normal PSU grade ones. They can be much better sounding as well as giving much lower punchier bass.
Last edited:
Hi Owen, thanks.
Found this bit of news so it seems Farnell is the only option although the prices seem to have been inflated.
Semelab ALFET - Farnell – High-performance audio transistors and high-reliability SiC diodes in stock | Electropages
Regards
Hi Owen, thanks.
Found this bit of news so it seems Farnell is the only option although the prices seem to have been inflated.
Semelab ALFET - Farnell – High-performance audio transistors and high-reliability SiC diodes in stock | Electropages
Regards
The heatsinks shown are actually slightly larger than what you mentioned. (Overall chassis size of 280w x 300d x 120h with 50mm heatsinks) I'm also hoping to squeeze a little extra heat exchange capacity by using finned 10mm plate for the lid and base, and coupling them as well as possible to the main sinks.
For my woofers (up to 600hz) I was only intending to run 100-150mA bias based off the graph shown in fig 8 here:
http://www.national.com/an/AN/AN-1850.pdf
Hopefully Owen will have time to check distortion vs frequency for a few different bias levels, so bias can be set according to the desired frequency range.
Another idea I had was to extend fins inside the chassis beside the amp modules. Essentially what I want to do is have as much fin area as possible
(within reason) in the smallest amount of space.
"ALFET audio transistors offer high performance, a small device footprint and excellent sonic characteristics for applications in the output stages of professional audio power amplifiers. Delivering reliable and robust performance when compared to bipolar solutions, the devices feature a wide safe operating area (SOA) and offer simplified protection, which saves cost and reduces design complexity."
This is talking about Mosfets vs. Bipolars, not Alfets vs. other mosfets.
Alfets are actually less well protected than mosfets such K1058, J162, etc, as the Alfets do not contain the built in gate protection diodes and so are static sensitive.
This is talking about Mosfets vs. Bipolars, not Alfets vs. other mosfets.
Alfets are actually less well protected than mosfets such K1058, J162, etc, as the Alfets do not contain the built in gate protection diodes and so are static sensitive.
thats what farnell do unfortunately, even worse here in oz than most other countries, can be >200% they then claim to provide 'free shipping' not all components, some are only marginally more, some even cheaper, but this type of premium device is often double or more. check out their thin film smd resistor or np0 capacitor pricing. one smd common mode choke i ordered (and cancelled) was nearly 1000% i have also caught them having the same item listed twice at 2 different prices and have brought this to their attention with no change
Last edited:
Hi
Excuse me if I intrude. Since the DPS-600 does not use large capacitors to store energy. (seen on pcb small capacity, just as a filter).
perhaps it can be helpful to show the behavior of the voltage rails with different types of signals, a MOSFET amplifier 450W rms AB. (see burst and continuous)
Regards
Roberto
Excuse me if I intrude. Since the DPS-600 does not use large capacitors to store energy. (seen on pcb small capacity, just as a filter).
perhaps it can be helpful to show the behavior of the voltage rails with different types of signals, a MOSFET amplifier 450W rms AB. (see burst and continuous)
Regards
Roberto
Attachments
I would be happy if someone repeats these steps with any type of psu, and obtained the same results (independent of the size of the psu and capacitors).. 🙂
AP2, do you know of any links to personal reviews from people using your SMPS in high quality / high end type amplifiers / full range Hi-Fi systems that you could post here?
I heard a Chord amplifier with what I was told was an SMPS a few years ago but it did sound a bit edgy / harsh on his B&W 801. Possibly not a fair assessment though, I think the amp was new and not run in.
I'd assume that such an SMPS would be a significantly lower cost than a big Tx and 60,000uF of good caps. And quite a bit smaller so also saving some costs on the amplifier case.
I was looking at the Spectron Musician amplifier last year. Spectron seem like they may well be the current kings of the digital amplifier, but when I looked they still used a 2kVA Tx and large bank of caps.
Thanks. 🙂
I heard a Chord amplifier with what I was told was an SMPS a few years ago but it did sound a bit edgy / harsh on his B&W 801. Possibly not a fair assessment though, I think the amp was new and not run in.
I'd assume that such an SMPS would be a significantly lower cost than a big Tx and 60,000uF of good caps. And quite a bit smaller so also saving some costs on the amplifier case.
I was looking at the Spectron Musician amplifier last year. Spectron seem like they may well be the current kings of the digital amplifier, but when I looked they still used a 2kVA Tx and large bank of caps.
Thanks. 🙂
Last edited:
AP2:
Thanks for the measurements! It's rare that anyone steps up with empirical evidence to prove performance.
I'd like to volunteer to buy one of these from you so I can run tests with this supply and a linear supply on the Audio Precision, and hopefully that will help alleviate some of the concerns people have with switch mode supplies.
I'll send you a PM and we can sort out shipping and availability.
IanAS:
The reason linear supplies like to have piles of capacitance on the output is because it's a brute-force way to compensate for the fact that there's no feedback, and hence no active regulation of the rails. A good SMPS, like the one shown in the measurements here, can have extremely good load regulation by virtue of the feedback used in the control system. If you performed the above tests on standard unregulated linear supply, the rail fluctuation would be immense, even with 60,000uF of capacitance.
You seem to be stuck on this notion that a whole bunch of capacitance can somehow make for better bass output, but that's just not the case. There is a lot more to the equation that just the amount of capacitance, and this is doubly true when you start looking at regulated supplies where the amount of capacitance means absolutely nothing compared to the regulation capabilities of the supply itself.
To put this into perspective, I've worked on an 800 watt SMPS designed specifically for a subwoofer that had a grand total of 4.4uF of capacitance in the form of two 2.2uF polypropylene caps on the output. It worked just fine, and performed significantly better than most large unregulated linear supplies. It's all about the design of the switcher itself, and the control loop. It has nothing to do with the amount of capacitance.
I'll do a head to head comparison between the linear and the SMPS, both measured and listening, and we'll see what happens.
Regards,
Owen
Thanks for the measurements! It's rare that anyone steps up with empirical evidence to prove performance.
I'd like to volunteer to buy one of these from you so I can run tests with this supply and a linear supply on the Audio Precision, and hopefully that will help alleviate some of the concerns people have with switch mode supplies.
I'll send you a PM and we can sort out shipping and availability.
IanAS:
The reason linear supplies like to have piles of capacitance on the output is because it's a brute-force way to compensate for the fact that there's no feedback, and hence no active regulation of the rails. A good SMPS, like the one shown in the measurements here, can have extremely good load regulation by virtue of the feedback used in the control system. If you performed the above tests on standard unregulated linear supply, the rail fluctuation would be immense, even with 60,000uF of capacitance.
You seem to be stuck on this notion that a whole bunch of capacitance can somehow make for better bass output, but that's just not the case. There is a lot more to the equation that just the amount of capacitance, and this is doubly true when you start looking at regulated supplies where the amount of capacitance means absolutely nothing compared to the regulation capabilities of the supply itself.
To put this into perspective, I've worked on an 800 watt SMPS designed specifically for a subwoofer that had a grand total of 4.4uF of capacitance in the form of two 2.2uF polypropylene caps on the output. It worked just fine, and performed significantly better than most large unregulated linear supplies. It's all about the design of the switcher itself, and the control loop. It has nothing to do with the amount of capacitance.
I'll do a head to head comparison between the linear and the SMPS, both measured and listening, and we'll see what happens.
Regards,
Owen
indeed, thanks heaps for stepping up to the plate; both opc and AP2!! i'll hold off on any psu related purchases till i see the results.
Hi,
Sorry, but I can not put the name of companies who are using this psu.
diy bought some and said they are happy, but I do not know how they put the forum post. one has in this forum. (quad404-AudioPower SMPS) (I know others, that will soon)
This SMPS has a very high cost of production, the components used and the manufactory (active power-unit and a special ultra-fast hybrid regulator).
The research on this new SMPS, is justified by the ratio size / performance obtained. This allows the creation of amplifiers with incredible realism of sound, stylish and ultra compact new design of case.
This SMPS has been designed for use in mono-block (one amp), then philosophy is that stereo is "dual mono". right way for a true audiophile.
----------------
I am happy that OPC head and publishing the results. AudioPower although, to date this has not pushed for DIY SMPS.
OPC .. I hope you have some very strong speakers. Perhaps no image, it produces a drop of just 400mV 600Hz to 20K form in terms of dynamics and timbre... 🙂
although I'm sure, that the first difference you will feel the right to 30-40w.
Yes, agree on feedback but i can not obtain this performances (response time and clean output) with traditional chip for smps pwm etc. then develop new,trafo is custom,etc..
Regards