Keantoken's CFP cap multiplier

As far as the power k-multipler, you are right with your reasoning. It is important that the reg doesn't have too good bass isolation, so that when the input dips, the reg can track it without saturating. However with the slower devices that will be needed, output impedance at HF will suffer. This doesn't include what may need to be done to stabilize it if it oscillates.

- keantoken
 
Okay, a less wimpy way of testing my own circuit is in the works.

Using a 2SC945P and the only PNP suitable power transistor I have, the 2SB772. Testing with 1N4148's to see if they explode at turn on (My guess is yes).

- keantoken
 

Attachments

  • DCP00885.JPG
    DCP00885.JPG
    291.2 KB · Views: 726

iko

Ex-Moderator
Joined 2008
You should stream live video! We all like fireworks :D

Edit: my guess is that it'll hold up just fine. On the other hand I just blew a couple of LEDs and a couple of MJE350 in my tube headphone amp prototype. But then, 20mA off can be pretty nasty at 270V :D
 
Only if I have live commentators!

Other prototypes temporarily displaced...

For some reason, my camera, an ancient Kodak DC210, has chosen to stop focusing right, among developing other strange problems. Good value for the money actually, considering how old it is...

- keantoken
 

Attachments

  • DCP00888.JPG
    DCP00888.JPG
    283.3 KB · Views: 709
kean,

From the initial simulation you posted I thought that the rolloff point was below 10Hz. If it is at 100Hz then I could use a capacitor larger than 1000uF for C1. I have run out of 1000uF so I am using 820uF. Would 2,200uF be OK? I have a couple left.

The ripples at the output of the rails look smooth, far from being jagged or toothy, and they are at 120Hz. So I see no chance of getting oscillation.

If I add another diode to the string, I expect the output will drop for another 0.56V or so. I can't afford this, as the existing Marantz reg requires some voltage drop to regulate (now given only 2.8V). currently, with the K-Muliplier in place the Marantz reg's input is 14.1V and output is 11.3V. The low voltage drop of 1.6V of the K-Multiplier makes its application possible. Fortunately, the load is constant current draw so things are predictable and the Marantz reg should not run out of headroom, given the output ripples of the K-Multiplier being only a few mV. All other circuits have no chance, unless I replace the Marantz reg completely.

So in conclusion, although the K-Multiplier has not reached its simulated result, it does provide some 30dB+ ripple rejection with low output impedance and low voltage drop, hence serves the Marantz very well. I am happy to live with it at its current state.

The only things I don't know what to do are - should I use BC550C/BC560C or the 2N5088/2N5401? The later had lower 120Hz ripples according to my previous measurements, but I am more concerned with high frequency performance than low frequency performance, and could I use 2,200uF for C1?

Regards,
Bill
 
Okay, a less wimpy way of testing my own circuit is in the works.

Using a 2SC945P and the only PNP suitable power transistor I have, the 2SB772. Testing with 1N4148's to see if they explode at turn on (My guess is yes).

- keantoken

kean,

I am happy to send you a full set of active device for the K-Multiplier if you can wait for a week or so for the parcel to arrive.

Bill

P.S. Silly me, since you have already soldered them. I am anxiously awaiting your measurement results.
 
Last edited:
The multiplier is up and running. Oscillating into a 1uF mylar film. Since your output filtering is so good, the multiplier may be oscillating it's heart out and you won't notice. The last thing to try I guess, is to remove any film caps you have at the output. I should have thought of this before.

You can use whichever transistors give the best result, but the 2N5401/5551 aren't low noise although they will work.

The output impedance bass rolloff is waaaaaaay down there, but the input isolation rolloff is about 100Hz for a 1000uf cap. Since rectified mains is nearly all 120Hz and above, it may not be worth increasing C1.

HiFi, that would be very kind, and would allow my to test out the regs on my headamp. However for prototyping purposes, I think this humble beginning is doing well enough.

- keantoken
 
Last edited:
Okay, I switched out the mylar output cap with a 1000u electrolytic, something with peach colored gibberish written down the side so it's got to be low-ESR. So far oscillations are undetected, or at least I can't distinguish them from what is usually propagating through my house.

Measured properly with coax with my 561B, output is rock solid, and shows >20uV pk-pk noise+anything else, with little drift. Slight ripples is visible above the noise. I think that's pretty good!

- keantoken
 
kean,

You reminded me that when I first tested it without any output caps I saw oscillation. But I thought the K-Multiplier may need some caps at the output for stability and I added some caps (small ones to start) and I never saw any oscillation again.

I don't have any film caps at the output, although 2 x 2,200uF ZL would give a very low impedance (0.0074R at 100kHz) which may be lower than many film caps.

Hand on, I have a faster way to get you the stuff.

Please send me a private email to both my work and personal addresses with the full list of all the parts you need, not just the 4 transistors. I can order them directly at Digikey and ask them to send the parts to you instead.

Regards,
Bill
 
Lol, current draw is... Zero.

I just wired up a 50 ohm resistor, will test with that, then measure ripples.

Okay, I just did that. Input ripple is 20mV, output is 50uV. That gives 400, which is ~52db input isolation. Still underwhelming considering simulation... But good nonetheless.

- keantoken
 
I have got your part list and am checking out an order from Mouser. It asked for a telephone number. I gave my Australia number but it rejects it. Can you email me a valid US number immediately?

P.S. and it is tricky - I have just entered Mouser's phone number and it rejects it too!

Again. I took out the leading 8 and it passed it. However, it says some items are back ordered, I am wondering if I will click on the continue button...
 
Last edited:
diyAudio Chief Moderator
Joined 2002
Paid Member
Lol, current draw is... Zero.

I just wired up a 50 ohm resistor, will test with that, then measure ripples.

Okay, I just did that. Input ripple is 20mV, output is 50uV. That gives 400, which is ~52db input isolation. Still underwhelming considering simulation... But good nonetheless.

- keantoken

From 58dB theoretical to 52dB practical I don't see anything underwhelming IMHO. That is about what the CFP does for filtering with practical parts it seems. Now if you wanted something like -60dB real, maybe you could make it RC, RC, instead of straight 1mF C1. If you want the low Zo and a super line regulation practical figure in some application, couldn't it work with two CCTs in a row?
 
O.K. Good luck on you.

The following parts will be sent to you tomorrow from Mouser:

2 x 2SA1930
2 x 2SC5171
10 x BC550C
10 x BC560C
20 x 1N4002
2 x 1k
2 x 100R
2 x Panasonic FC 1000uF 25V

There is a back order of 20 x mpsh81 that will be sent to you separately. These are for my use for the future rebuild of the Iko regs on power amp chassis. Since the shipping costs to Australia starts from $30 it would be kind of you to send these in a normal envolope to me, which should cost you perhaps a couple of dollars, saving me $30 in shipping, if you don't mind.
 
Great! Thanks a bunch.

DIYAudio is doing weird things, and I don't see Iko's reply on the forum, but I got the Email.

Iko's circuit has much better input isolation, because of the cascoded transistors, which eliminates the detrimental Early effect. I tried to come up with a reliable meld between this and the CFP multiplier in the simulator, to combine the virtues of the two, but I wasn't successful, at least not yet. The problem was more robustness and reliability than whether the circuit actually worked. In any case, adding another transistor will almost surely add another 1.3V drop.

The circuits are different, but the goals are related, so I would say discussing them both here may be helpful.

- keantoken