My take on a discrete shunt voltage regulator

[iko]

Heeeeey! That's the spirit!

But where are we going to get this kind of crowd:

http://www.youtube.com/watch?v=28GaKoCuobU

or this

http://www.youtube.com/watch?v=Zbj0iyI4NMQ

or this

http://www.youtube.com/watch?v=TZ_fUdI-puE

... I could go on

 

[keantoken]

I'm subscribed, so even if I don't post, I always notice when a post is made, usually to quickly read through it.

How is the noise amp coming along? I'm really looking forward to the spectrum measurements.

I had an odd thought. LTSpice doesn't take into account real time temperature. If temperature has a large enough affect on output, we may actually find that in real life LF/DC impedance is much higher, or even that there is a negative impedance at DC! IE, temperature variations caused by load changing DC output. I wonder if just the MOSFET getting hot is enough to do this. I wonder if this dynamic is audible?

- keantoken

 

[iko]

I finished a notch filter today and was testing it, when all of a sudden when walking over a bridge that takes me to the electrical engineering dept. building I saw an old oscilloscope out at the side of the building. The EE people get rid of old stuff at times. So there went all of my free time trying to resurrect this old HP 140A scope.

Regarding the temperature effect on the regulator. I know this for sure, and it was very obvious on 5d. Good lesson there. The mosfet will increase its bias as the temperature goes up. When Vgs reaches a certain point the higher temperature (device dependent) the current will decrease with temperature, but before that happens, you get nice temperature run-off. I don't know and don't want to speculate about sound and such.

I put 5k through the spectrum analyzer, did you miss that? But yes, I need to amplify the noise because at the moment the spectrum analyzer internal noise is masking the regulator noise. I'd still vote the Jung SR for lower noise. We're not using fancy Vref devices. Perhaps we have a chance at lower output impedance, but that has proven so far a real pain to measure (for me).

Slowly, we're getting there...

BTW, I'm really looking forward putting your HP amp through the spectrum analyzer

 

[keantoken]

Awww, too bad I don't live near any electronics places. There's no good junk for miles around, barring what can be found in dumpsters. And all the people here are super frugal.

I didn't realize you could test my amp with the analyzer... I hope simulation is close to reality (I don't see why not). Ultra-cool. I don't know when I'll get back to my own prototype, since I recently had the opportunity to grab some lumber for my damped open baffle project...

I saw an analyzer shot but it was just noise floor all across. Maybe I just don't know how to read it.

- keantoken

 

[iko]

Awww, too bad I don't live near any electronics places. There's no good junk for miles around, barring what can be found in dumpsters. And all the people here are super frugal.

Yeah, I hear you, I can only imagine what stuff can be found near military bases. This is a university, so only low key stuff is thrown out sometimes. I'm not complaining, some good finds so far.

I didn't realize you could test my amp with the analyzer... I hope simulation is close to reality (I don't see why not). Ultra-cool. I don't know when I'll get back to my own prototype, since I recently had the opportunity to grab some lumber for my damped open baffle project...

Sounds to me like projectitis. I have that disease in fact

I saw an analyzer shot but it was just noise floor all across. Maybe I just don't know how to read it.
- keantoken

You're not mistaken; the regulator noise was below the noise floor of the analyzer. With the low noise amp more will be shown, but why do I even care? I'm really not that obsessed with low noise. Plus, Bill from AU tested it with his ears.

 

[AndrewT]

Hi Iko,
little did you realise on 21st October that you would have generated all this interest and cause so much hassle for the retailers' buyers.

 

[jackinnj]

Speaking of notch filters -- here's a link to the Hewlett Packard Journal in which their digitally controlled notch filter for the 8903A is discussed:
http://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1980-08.pdf

 

[HiFiNutNut]

Oh come on! You can't be serious! Here's a test.

Hey, guys, if you're reading this, just say hello. Why not, a simple hi there!

Watch this space: Film | Digi-Key

There are only 385 left now. Two days ago I saw 715. I bought 100. Then I posted on this thread. At noon 100 was gone. It was on the Easter holidays so no manufacturers would have bought them. Yesterday, 110 was gone. Today another 130 was gone.

I bought a few of those 1.1uF MKP a while ago and found them good. So last time when I was accumulating my list of parts to be ordered from Digikey (for the purpose of saving freight by buying in a batch) I watched the stock of that particular item at Digi-key carefully over a few months period. When the stock was down to about 200 about 6 months ago, I bought 2 dozens. I recalled they went a lot slower than this time. They have been out of stock for possibly 4 months and Digikey must have re-stocked them recently.

Hellooooooooooooo? Who are building these regs? Say Hello!

 

[iko]

Nice link jackinnj!

Bill, I might get some of those caps myself, I'm persuaded

 

[massimo]

Aren't the 1uF/63V MK416 as good as the 1.1uF?
Over 4,000 pcs available......

P.S. ordering 1,200 pcs the price drops by nearly 60%...... only 12 DIYers or so needed

 

[HiFiNutNut]

Aren't the 1uF/63V MK416 as good as the 1.1uF?
Over 4,000 pcs available......

They are as good. I have got a few of them too. Getting the largest value is for space saving. The largest value in the same series manufactured is 1.2uF. But I have never seen them available. Second largest is 1.1uF.

Since these are good for bypassing and coupling, i.e. rail output, Vref for the regs, output of CD player, shunt for electrolytic capacitors, etc, the actual value and precision is not important. But the larger value you get, the more space you save for a given capacitance. If you compare the size of these MKP to MKT of the same voltage, you will find the size (volume) of the MKP may be about 10 times larger. Dialectric aborption, i.e. the ability to charge and discharge quickly without "hanging-over" the signal, of MKP is better than MKT. This is probably the reason why polypropylene and polystyrene sound a lot better than others, except perhaps Teflon, that is far more expensive and even much larger in size. See Jung's article for capacitor tests. That is the only article I trust, many others are too subjective in my opinions.

Please be careful with the voltages of these capacitors. The one I recommended was 1.1uF 63V, possibly the smallest size of MKP you can find, making it practical for PCB use. However, they are rated 63VDC and 25VAC. You may not be able to use them for passive crossover, for example, because the AC voltages likely exceed 25V. For the regulator use, they should be fine provided your rails are lower than 63VDC.

 

[klewis]

Oh come on! You can't be serious! Here's a test.

Hey, guys, if you're reading this, just say hello. Why not, a simple hi there!

Hello, this is my daily (ok sometimes hourly) esoteric fix...

 

[TheShaman]

Hi! Keep up the good work!

 

[iko]

Hey, glad to be entertaining.

Here's something to ponder. Any friend of google who looked for regulators must have come across the Wenzel finesse voltage regulator. Yes, it's as old as Nefertiti but never embalmed. Do not take this lightly, because these are serious people, with patents. So go and read about the Wenzel finesse voltage regulator at this link.

Finesse Voltage Regulator Noise!

Recently this kind of regulator has been used by top engineers in certain restricted venues.

My first attachment is the finesse regulator next to a noname LC low pass filter. The LC low pass filter is sloppily built, with magnet wire that sucks (150 ohms series resistance).

The second attachment is the simulated frequency response of both circuits. V(x) is the finesse, and V is the LC filter. Up to 10MHz they are almost identical in psrr.

"Hey!" -- you'll say -- "Buddy! -- look beyond 10MHz, the finesse reg keeps on decreasing psrr. Your stupid LC filter sucks beyond 10MHz."

Too true... only too true...

 

[iko]

"Hey!" -- you'll say -- "Buddy! -- look beyond 10MHz, the finesse reg keeps on decreasing psrr. Your stupid LC filter sucks beyond 10MHz."

Too true... only too true...

Alright, the LC filter sucks beyond 10Mhz when compared to the finesse reg? What to do? Let's replace the cap with a resistor. It's cheaper anyway. V(x) is again the finesse regulator psrr, V(z) is the LR low-pass filter psrr. It's difficult to see any difference, really, as they're almost identical.

Yet, the finesse regulator exists and is preferred to a simple LR filter. Does any of you know why that is?

 

[Salas]

Did you account for parasitic capacitance of the inductor?

 

[iko]

Noooo, it's a perfect choke What do you suggest?

But then again, I didn't add trace and other parasitic properties of the finesse regulator either.

 

[keantoken]

Are you sure the opamp will perform as well as simulation suggests?

I'm wondering if a discrete version of the Finesse (if it's really worth it), would perform better and have higher BW.

- keantoken

 

[Salas]

If it has a core its much more than an air core. They wouldn't use a speaker coil on a PCB due to cost and bulk easily. Then again a speaker coil would have many layers. Say 25pF.

 

[iko]

I'm not sure kt. Salas has a very good point, add all the parasitic properties to the circuit and watch it go up in flames. Both the passive and the active filter.