Interested in ADM7151 regulator?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
There are input/output caps on the regulator module, so the ground pin of the 3 terminal connection will carry AC currents from the PSU ripple and load current,. This will corrupt the ground reference, and the performance of this regulator will not be better than a LM317 (it could actually be worse), unless the load draws a constant current and the incoming power supply is already very well filtered, in which case, one has to wonder why using an expensive regulator with such high PSRR and low noise...



Yes. Decoupling caps are always located electrically close to the load... the inductance of a properly mounted hand-solderable SMD ceramic is lower than 1nH. The black-box POSCAPs that Coris used, again when properly mounted, are about 1.5 to 2.5 nH, that's equivalent to a wire loop of about 1 to 1.5 mm2 area... putting them on a daughterboard with a 3 pin connection will add 10-50x more inductance, it is a waste of a good capacitor, you could use a noname electrolytic, the load wouldn't see any difference.

The regulator module should only have the reference bypass caps, so that almost no current flows in the ground pin. If this makes the regulator unstable, you need shorter connections.

If you want to extract the full performance of this reg while putting it on a daughterboard, you will need at least 2 ground pins, one for current and one for sensing. 2 output pins (current and sense) are also a plus.

So what is the solution you suggest? We have as a given that we can't have on tha same board the dac or the target chip *and* the reg. What do we do?
 
Just like stormsonic's little drawing... the problem to avoid is having a common impedance (here, the GND pin) shared between decoupling capacitors (which carry noise currents) and reference ground (which should be clean).

1) audit the decoupling on the existing board

If the decoupling sucks, check if you can add more caps on the DAC board.

If it is sufficient, then put the regulator and its reference filter cap on a mini daughterboard with very short connections, as short as possible. Here you got your 3 terminal regulator.

Now where to put the input and output caps ?

For example you can scrape the soldermask on the ground plane around the 3-terminal regulator connector and solder the input and/or output caps there. That way it will not share a common ground with the reg's reference. Both cap's GND and the reg's GND are connected to the plane, which has much lower impedance than a O.1" header pin.

If it is stable and there is enough decoupling caps on the DAC board already... no need to add more.

If it is unstable, an output cap will be needed... Try a cap with enough ESR to prevent creation of a LC resonant tank with the other caps on the DAC board. For example a 20-30 mOhm ESR POSCAP as Coris used should do nicely.

Next week I'll try to make and measure some CLC filters for my power supplies using ferrite beads, I'll keep you posted. I'll also measure the inductance of the 0.1" header ;)
 
Member
Joined 2009
Paid Member
peufeu there is all right. You explain well, and you have right in the most of cases. Nothing to doubt about. But I wonder on something...

As you can see, I and Nikitas have each on his own elaborated a design to make this new regulator chip available to those interested to try it. I do not state that my design is the best, but it is functional. The things may be always done better.
I see you know very well in depth the details about how to be designed a regulator which it may use this chip. Of course a such PCB is not very complicated to be designed.
I just wonder why you can not come out with a so perfect design for this kind of regulator, or for something else, showing us in fact how the PCB it have to be designed, how it may measure your device, and so on?

I see you wonder why using so expensive and high end regulator chip, when one can get the same performances from a 317 one with the right design, decoupling, filtering and so on. Can you please produce a such design (kit/module) and show us that your 317 regulator perform better than this ADM7151 in these two design versions available today?
I can assure you that I will buy it...
 
Last edited:
> I see you wonder why using so expensive and high end regulator chip,
> when one can get the same performances from a 317 one with the
> right design

There's a misunderstanding here, of course an old POS like LM317 can't perform as good as this regulator. LM317 can't work with ceramic caps anyway, or anything low-ESR, so it would not work at all on a load that desperately needs HF decoupling (like a DAC).

I just said that this high-performance regulator will give no better (even worse) results than a good old LM317 if its reference ground is corrupted.

Simulation example below, I used reasonable values for pin contact R and L. You can see the PSRR is no good. One of the main selling points of this chip is its excellent PSRR...

Solution is just to remove the capacitors on the regulator module and solder them on the board (and only if they are really needed).

Better solution, of course is to use more pins. I use a 2x9 pin connector with 2 IN, 2 OUT, and 14 ground pins.

I just wonder why you can not come out with a so perfect design for this kind of regulator, or for something else, showing us in fact how the PCB it have to be designed, how it may measure your device, and so on?

In fact I'd like to measure the impedance of this one, so I'll add it to my next mouser order... but this isn't going to happen tomorrow since I always wait until the list grows over the free shipping threshold!...

In the meantime, if you got a TL431 and some transistors you can always try this one :

http://www.diyaudio.com/forums/solid-state/253390-experimentations-regulators-3.html#post3879434

But, it will not work either in 3-terminal format since it is a shunt, and that means it has lots of current to dump in the GND. Only a series reg with very low GND current will work in 3 terminal format.
 

Attachments

  • PSRR.PNG
    PSRR.PNG
    12.3 KB · Views: 680
  • sim.PNG
    sim.PNG
    21.2 KB · Views: 678
Last edited:
> I see you wonder why using so expensive and high end regulator chip,
> when one can get the same performances from a 317 one with the
> right design

There's a misunderstanding here, of course an old POS like LM317 can't perform as good as this regulator. LM317 can't work with ceramic caps anyway, or anything low-ESR, so it would not work at all on a load that desperately needs HF decoupling (like a DAC).

I just said that this high-performance regulator will give no better (even worse) results than a good old LM317 if its reference ground is corrupted.

Simulation example below, I used reasonable values for pin contact R and L. You can see the PSRR is no good. One of the main selling points of this chip is its excellent PSRR...

Solution is just to remove the capacitors on the regulator module and solder them on the board (and only if they are really needed).

Better solution, of course is to use more pins. I use a 2x9 pin connector with 2 IN, 2 OUT, and 14 ground pins.



In fact I'd like to measure the impedance of this one, so I'll add it to my next mouser order... but this isn't going to happen tomorrow since I always wait until the list grows over the free shipping threshold!...

In the meantime, if you got a TL431 and some transistors you can always try this one :

http://www.diyaudio.com/forums/solid-state/253390-experimentations-regulators-3.html#post3879434

But, it will not work either in 3-terminal format since it is a shunt, and that means it has lots of current to dump in the GND. Only a series reg with very low GND current will work in 3 terminal format.

Thank you for all of your helpful advice :)
 
Hi guys,

I had some personal matters to deal with but finally I'm back :)
I received today parts from mouser but I am waiting for the pcbs...
For those who aren't interested in using the 3pin header I added 2 terminal block positions.
Or you could solder wires directly on the pcb with separate in+gnd and out+gnd.

SMD pads are oversized for easy soldering-desoldering.

I also got some caps+inductors for the filter i'm planning to make for the smps psu before the ger board.
 

Attachments

  • IMAG0718.jpg
    IMAG0718.jpg
    378.8 KB · Views: 553
  • bottom.jpg
    bottom.jpg
    65.4 KB · Views: 530
  • top.jpg
    top.jpg
    82.4 KB · Views: 521
No problem, I will check em today if they work well and then I'll let you know!

By the way, since this is the first version it would be nice if you send me feedback with possible refinements,additions etc when you get them :)

A couple of things I did last minute before production was to add positions for terminal blocks ini addition to the 3pin header and make the pads bigger so we can easily solder/de-solder parts for experimentation.
 
Thats great Nikitas.
I will be comparing these to the TPS7A4700 Regs which I already have to see whether they are audibly better and provide some feedback. Unfortunately my oscilloscope is quite basic and noisy so noise measurements are a bit of a problem at this kind of level. The layout of your board seems not to contradict any of the principles in the AD Evaluation Board so looks great to me. I can have a bit of a play with component selection and values though.
Thanks for your efforts in creating this board!
 
Thats great Nikitas.
I will be comparing these to the TPS7A4700 Regs which I already have to see whether they are audibly better and provide some feedback. Unfortunately my oscilloscope is quite basic and noisy so noise measurements are a bit of a problem at this kind of level. The layout of your board seems not to contradict any of the principles in the AD Evaluation Board so looks great to me. I can have a bit of a play with component selection and values though.
Thanks for your efforts in creating this board!

Yeah, I followed the instructions given by AD's datasheet and took some advice from articles I have read about the types of components.
 
Everything works well! Me super happy!:D:nod::cheers:

I had it hooked for more than 30 minutes at 5v with the indication LED on and a resistor-load. No signs of heat whatsoever. Tried voltage range from 1,4 to 5,2v.

I even pushed to 5,4V for about 2 minutes, but I want to test it further for stability and heat for anything above 5,2v.

But for 1,4 to 5,2v works perfectly well!!

During the weekend I will also test it in my application: 9v wall adapter and a custom LCL filter ADM7151 to see what happens.
 

Attachments

  • 2014-05-16 23.34.42.jpg
    2014-05-16 23.34.42.jpg
    347.7 KB · Views: 605
Nik
Just had another look at your board design. I understand that the low frequency performance of this regulator is very much dependant upon the value of C Byp, see attached graph (courtesy Hifiduino).
http://hifiduino.files.wordpress.com/2014/02/adm7150by.jpg
In your next iteration, it might be a good idea to allow space/terms for say a 1000uF electrolytic in parallel with the ceramic and the tantalum?
I know the Datasheet recommends solid tantalum but to get up to these kind of values would consume a lot of space (needing multiples) and also would be very expensive?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.