Progress on my Amplifier

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hey guys, I've made some major headway on my amplifier project. It uses three Aussie amplifier modules, the NX150's in a DIY Enclosures Chassis, and power supply components from ApexJr. The Power supply is housed in a seperate chassis and isn't finished yet, though I have pictures of it in test setup. It uses a CLC construction with 2x600va toroidals, followed by a 22,000uf cap per rail, then an 11mh inductor, and then is sent off to the amplifier chassis, which as another 22,000uf cap per rail. In the future, I intend to build three indipendant power supplies, which I have the parts for, just not the room. I couldn't figure a way to fit 6 caps, 6 Inductors, 3 transformers, and numberous other associated parts into the one power supply chassis, let alone another 6 caps in the amp chassis. Anyway here are some pics.

612760684_e0a13e2575_b.jpg

612760620_511b49b1f3_b.jpg

612760600_af1b264e7f_b.jpg
612760582_aa1ca841c3_b.jpg
 
First, I'm using such large chokes because I wanted to use a CLC power supply, and thats how big the appropriate sized chokes were. As for the caps, well, I got a good deal on them. I couldn't beat that price with smaller units. Also, finding caps rated at 100 volts and say 10,000uf's isn't easy, choices are limited. These are 22,000uf's and 100 volts, so you aren't going to find the same value in say half the package, but something like the Panasonic TS could have worked and would have been considerably smaller. However I would have needed twice as many of them, and I believe cost would have been around 15 dollars each for those. I paid 15 dollars each for these, but was able to use half as many. They also have a lower ESR than the Panasonic TC's would have, though I don't know if 2 or 3 wired together would have been lower than my one 22mf cap.

Supposedly these transformers have a built in softstart function. I have no idea how or what that means, that is simply what Plitron claims. I will say that in testing I didn't notice any light dimming with the one 600va tranny, but I did notice it with my 350va chinese sourced toroidal. Who knows, maybe there is something to that. However, if I do get tripped breakers or light dimming, I will be getting a softstart for it, and trying that.
 
Hi,
your second stage caps are rather small. How's that for the opposite opinion?.
Three channels fed from +-22mF is asking a lot.
I reckon that a single channel driving an 8ohm speaker needs +-15mF to +-20mF. For three channels fed off a common supply you are in the range +-45mF to +-60mF. If you want to drive 4ohm speakers you should double all those capacitance values.

BTW,
all the peak current into the amplifier(s) come(s) from the last stage of smoothing. The first stage provides a rough DC supply to the inductor but the inductor prevents fast recharge of the second stage and prevents effective low impedance supplies to the amplifier. The last stage provides the low impedance supplies for the amplifiers.
Have you used PSUD11 to check ripple in the first stage smoothing? Your 22mF caps are probably OK, but still worth checking.
 
Yes I have checked the supplies, and actually I was reconsidering the design just now while at work. I have more than enough caps for this, and was thinking about how I could move things around to handle a few more caps. I think there really isn't enough room in the amp chassis, but I could put 2 more caps in the supply chassis. Then I would have a 22mf cap then inductor, then 22mf cap, then power wire, then 22mf cap. I believe this will essentially create a CLCRC supply with a very small R value, probably half an ohm or less. None the less, it does nothing bad for the performance.

Now as for numbers, I ran it every which way to see, and basically this is the numbers at peak draw. Using a supply load of 10ohms it draws say around 6 amps. At the load the output voltage is 59 volts and the ripple is 10mv. Ripple current is 1.05ma. At C1 the ripple is 2 volts, with a range between 58 and 60 volts. Ripple current at C1 is 51 amps...??? Is that right, 51 amps ripple current at C1? Oh wait, because its in parallel that makes sense now, I think. I mean, 2 volts of ripple at the first cap is a lot, but I think its within the accepted range. Increasing the first state of capacitance to 44mf only changed it to 1.85 volts difference, so I don't see it being worth my while.
 
Ok now I'm a little confused, so Andrew if you can help on this, I would appreciate it. The Ripple current for the Hitachi caps is a max of 9.7 amps rms at 120hz. Now I'm getting ripple current at C1 of 54, and an RMS of 14, does that mean at peak draw I'm exceeding the ripple current rating of the caps? The caps have a rated ESR of 7mOhms, and an ESL of 30 mH's. I didn't even think to look at ripple current at the first cap, I was paying more attention to ripple voltage at that cap.
 
Hi,
that is the problem with rCRC and rCLC.
The first stage carries enormous ripple and why I directed you to check.
If I were doing this I would use many small cheap caps in parallel in the the first stage (cheap and any quality since they are too removed to affect sound quality).
Then place high quality low ESR caps NEXT to the amplifier PCB to satisfy the peak current demand. Your extra Cr (=rCLCrC) will reduce ripple but in my opinion do nothing to help the first or last stage. Use the extra caps on one of your PCBs. If you were to place film decoupling on the PCBs you may be able to locate the large electrolytics a small distance away from the amps (similar to your first pic).
I would also place +-22mF on each amplifier rather than a common power bank.

Why did you choose 10ohm test/simulation load?
Three channels running simultaneously could be much worse.
 
I tried various ones, and 10ohms simply gave the most accurate current draw to what my transformer could give. I tried 5ohms which gave 11 amps draw I believe, and it didn't change the ripple current much. Now I also tried removing the LC state and made it just a capacitor filter, and I still got 15amps of ripple current. Scratch that, with 5ohms I get 25 amps of ripple current, rms. I must have done something wrong with my last simulation. However, removing the LC still showed the same level of ripple current on that capacitor.

I would prefer to use these caps because they are what I have available right now. Using smaller cheap caps would increase my cost some as I would have to order those all over again, and then find a new way to mount them. It's not the end of the world, but still something I need to consider.

The amps have 100uf's on the supply lines onboard, with film bypass caps, wima .01uf caps. There is also various stages of decoupling with film caps in other areas. I don't think I have enough room to place 6 caps in the chassis, in fact, I know I don't. Given that these modules are rated at only 150 watts, it seems a little crazy to give each one 44mf of capacitance, thats 120mf's for 450 watts. I'm not saying it wouldn't be nice, but that is far in excess of any production Class AB amps I know of. Just looking at something like say the Parasound Halo amp line, they use on their similar rated 5 channel amp 1.4kva transformer and 100mf's total, thats for 5 channels, or 750 watts. I have 1.2kva and 88mf's of total capacitance for 450 watts. Plus I have the inductors, though I'm starting to see your point about the increased load on C1, I thought using a CLC allowed less capacitors to be used.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.