Any experience powering opamps from DC-DC converter?

[BJosephs]

Hi All,


I have a project that needs two ground-referenced opamps added but I have no useful power supply rails to work from (it's a tube amp). I can get 5-8vdc supply with substantial current and I was thinking of using something like this:

NDTD0515C Murata Power Solutions | Mouser

To generate a bipolar supply. I've never used such a device and I was hoping someone has and can let me know the caveats relative to high-quality analog audio. I noticed the minimum current requirement to maintain ripple spec. I can also use linear regulators after the converter if need be. I'm trying to avoid adding a second, albeit small, power transformer.


Thanks,

Brian

Sent from my SM-N900V using Tapatalk

 

[stratus46]

I didn't use Murata but did use Computer Products DC-DC converters to power opamps in a low noise HDTV video preamp in a flying spot scanner. They worked great and avoided ground loop problems. The boss used these converters in other units as well and were chosen because of low ripple noise

G²

 

[Calvin]

Hi,

it works excellent, imho better than most any other option.
Look also for the NMK Dual output series and the MEE1 Single output series.
The latter offering outstanding values for regulation and ripple values (even beeing non-regulated types the regulation is very good)
You may series connect the secondaries for Dual outputs.
Adding some post-filtering and modern LDO like TI´s TPS7A series or LTs fantastic LT3042 and You don´t need to worry about OPAmp supply line qualities.

jauu
Calvin

 

[jean-paul]

The old fashioned "small toroid-rectifier-caps-regs" solution might just outperform most switchers easily....

I have used MCFM32/12 for this purpose with good results. If you need to use linear regulators after a switcher what is the point of using a switcher ? If you decide to use one you will need to shield it to avoid it straying in garbage in the tubes. You don't want to hardwire the switcher so you need to design a PCB anyway whilst PCBs for symmetrical PSUs can be found for little money. IMO using a switcher in an amplifier is asking for extra work. It is already an art to make a good clean PSU for 50 Hz let alone one that runs on 100 kHz....It is more in line with the purpose and philosophy of a tube amplifier (and their owners/repairmen) to keep things simple, reliable and as clean as possible. Not sure but hooking up a switcher to the filament supply also might pollute the filament power with distortion/harmonics/RF ! It is somewhat contradictory to use tons of semiconductors running on higher frequencies in a 4 tube amplifier

Symmetrical PSUs:

http://www.ebay.co.uk/itm/78xx-79xx...AeeDlqzwcraxbL9%2Fpzg%3D&orig_cvip=true&rt=nc

http://www.ebay.com/itm/DIY-PCB-Bipolar-opamp-power-supply-/151746532864

http://www.diyinhk.com/shop/audio-k...supply-linear-regulator-dual-12v-150max2.html

 

[jean-paul]

Nicer and quite compact one:

4.17uV Ultralow noise DAC power supply regulator +-12/15V 1A - DIYINHK

You did not mention if the opamps are used in the signal path. If they are used for biasing tubes you might have other design issues to take into account.

 

[thoglette]

Short answer: my experience is they create buckets of HF hash
Long answer: it's never quite that simple, but if you need low weight and/or high efficiency they can work.

 

[Calvin]

Hi,

It is already an art to make a good clean PSU for 50 Hz let alone one that runs on 100 kHz....It is more in line with the purpose and philosophy of Tube amps...

Yes, that precisely names it .... philosophical views
Rectifying 50Hz or 60Hz creates NF-garbage in a freq-range where the human hearing is most sensitive ... why should that be desirable or preferrable against a power source that creates only some garbage in the ultrasonic range, where it can be dealt with by high quality parts with compact layout and build?
And we're talking of rms values in the low mV-range up to 20MHz here.
Power line frequency rectyfying instead requires bulky parts with large values if the ripple values shall be low and as the bandwidth frequency limit of the filtering must be very low too.
Due to the low refresh cycle the transient currents drawn through the rectifiers are comparatively large (accompanied by hf-hash).
The switchers may be used with passive post filtering only already, the same way a transformer based PS with capacitive filtering may suffice, but wo question does LDO post filtering improve matters in both cases.
I'd regard a switcher rather as a alternative/replacement for a transformer/rectifier/caps power supply.

jauu
Calvin

 

[jean-paul]

Hi,

I'd regard a switcher rather as a alternative/replacement for a transformer/rectifier/caps power supply.

jauu
Calvin

Exactly !

AFAIK the switcher also needs a rectifier and cap as it is fed from the filament windings....as OP stated "I can get 5-8V DC supply with substantial current". For the two opamps that are mentioned probably 2 x 12 or 15V is needed @50 mA max. so no large caps and large transient currents... but merely a toroid the size of a walnut and a tiny regulator PCB

- Practically it seems only the small mains transformer is an obstacle for the OP. Maybe the OP can elaborate why the transformer is such an obstacle ?!

- We also don't know if the opamps are used for audio or other uses. If they need to do a non audio function even 2 x 6V can easily be created with the filament winding.

BTW I never encountered a device that did not suffer in sound quality when RF was present. For instance devices that have Bluetooth are notorious in that regard.

 

[BJosephs]

Thank, everyone. Certainly a full range of opinions here. I appreciate the thoughtful responses. I'll think on this a bit more.


Brian

 

[jean-paul]

It would help a lot if you would give feedback by replying to questions as we don't know some things which might lead to wrong advice from our part.

Just one post before yours there are some questions ...Communication goes 2 ways in general.....

 

[JMFahey]

Hi All,


I have a project that needs two ground-referenced opamps added but I have no useful power supply rails to work from (it's a tube amp). I can get 5-8vdc supply with substantial current and I was thinking of using something like this:

NDTD0515C Murata Power Solutions | Mouser

To generate a bipolar supply. I've never used such a device and I was hoping someone has and can let me know the caveats relative to high-quality analog audio. I noticed the minimum current requirement to maintain ripple spec. I can also use linear regulators after the converter if need be. I'm trying to avoid adding a second, albeit small, power transformer.


Thanks,

Brian

Sent from my SM-N900V using Tapatalk

1) Op Amps have very good supply rejection to begin with.
Of course, a little extra care filtering +/-15V rails always helps to take any remaining ripple/hash/noise even further away.

2) I regularly use switchers to get +/-12V from 12V batteries in my portable MI amplifiers, never a problem.

3) Murphy's Law at work, Mouser is down for maintenance so can't check yours, but no reason to believe it won't work.
Just add extra 100r+100uF passive filtering to each output as a precaution , if you wish add a 100nF ceramic across the electroytic.
More important is how/where you ground that extra supply, try a couple ways to find best.

4) slightly worried about your " 5-8vdc supply with substantial current ", I bet you count on rectifying filaments ... big problem is that you need one end grounded but *both* are 3.15VAC away from ground, because with tubes you ground the center tap, either real or virtual.
That requires your switcher to work "floating" and have transformer insulated outputs, even if ferrite transformers.

5) that said, if your power transformer has a 40 to 50VAC bias tap like many Fender amps do, you can easily feed your Op Amps from there (properly rectified and filtered, of course).

6) it would help if you posted a schematic of what are you trying to build.

 

[BJosephs]

It would help a lot if you would give feedback by replying to questions as we don't know some things which might lead to wrong advice from our part.

Just one post before yours there are some questions ...Communication goes 2 ways in general.....

Apologies, there were lots of opinions and I wanted some time to think them over before replying, but also without going radio-silent. I had the reply window open for so long this morning your last post came in before I refreshed.


So, I have a tube amp design that sounds great but has poor input sensitivity, requiring 10Vp-p at the input for full output. There is a thread here:

http://www.diyaudio.com/forums/tubes-valves/288853-different-type-el84-se.html

I could just add another tube stage but up until now everything I build up changes the sound too much. So, I started to consider adding two simple fixed gain-stage opamps to raise the signal by 5x to 10x right at the input. I usually buy the Express PCB miniboards which are 2.5"x3.8" fixed size, so I'd like to pack the whole thing onto one such board.

- Practically it seems only the small mains transformer is an obstacle for the OP. Maybe the OP can elaborate why the transformer is such an obstacle ?!

jean-paul: Just for size/space reasons. I have shopped around for small toroids like 1.6 and 3.2VA encapsulated unit and even they are 40mm square and take up too much of my PCB real-estate. I also didn't like the idea of bringing 120vAC right to the board... If I go linear then I think your recommendation of MCFM32/12 or similar would be the best. I could just find some empty chassis space and bolt one of those down, then just put rectification and regulation/filtering on the board with the audio stuff. Also, thanks for the links to the PS boards.

4) slightly worried about your " 5-8vdc supply with substantial current ", I bet you count on rectifying filaments ...

JMFahey: Yes, I was thinking about rectifying the the heater supply but as you point out I will have a ground issue. I want to keep the heaters differential for hum suppression so I guess that plan is out. Thanks for saving me some trouble there. I don't have any other taps available, sadly. Thanks confirming that you've had success with switchers - this comes up for me every once in a while and I've never taken the plunge.

it works excellent, imho better than most any other option.

They worked great and avoided ground loop problems.

stratus46, Calvin: Thanks for the confirmation that this can be a workable solution. Like I said above, I've always wondered if this would be acceptable for audio.

In the end, the issue with getting the 5vDC from the heaters really disqualifies the DC-DC converter idea for now.

 

[Mooly]

I've kept looking at this but haven't commented...

So just a thought. You need a gain of up to 10 (which is easy for an opamp stage) and you don't want to compromise grounding and so on.

A very workable alternative could be a simple AC coupled opamp stage resistively fed from the main HT line together with a simple zener stabiliser. You should be able to get away with a single 5 watt or couple of 2.5 watt series feed resistors.

 

[jean-paul]

A very workable alternative could be a simple AC coupled opamp stage resistively fed from the main HT line together with a simple zener stabiliser. You should be able to get away with a single 5 watt or couple of 2.5 watt series feed resistors.

"able to get away with"... ----> We better design stuff well and swim against the stream.

This is the worst plan ever to feed opamps. Suppose what happens when the Zener diode breaks down... Outright dangerous !!!!

So the power amp needs 10Vp-p for full output swing. That is simply a bad design. An amplifier should ..eh.. amplify...Anyway it should amplifly a line signal enough for full output swing. This design does not. A problem is best solved at its root.

OP might better use a high mu tube in the first stage....and forget about the opamps. Tube amps are simple and high mu tubes are easy to find. The KISS principle ... The opamps will determine the sound character anyway and seem superfluous in this application. The switcher won't make matters better. It all seems a very long road to good results.

If opamps are allowed it is possible to just use opamps with one of their best properties (high gain) to drive the power tubes and leave out the driver tube. Kind of a hybrid and possibly best of both worlds. There will be enough space for the small toroid

jean-paul: Just for size/space reasons. I have shopped around for small toroids like 1.6 and 3.2VA encapsulated unit and even they are 40mm square and take up too much of my PCB real-estate. I also didn't like the idea of bringing 120vAC right to the board...

The same transformers are available "naked". Those like MCFM32/12 are meant to be bolted down to a chassis with just one bolt. 120 V AC will not be in the vicinity of the audio boards when engineered right.

 

[infinia]

interstage signal transformer? I reckon the turns ratio in reverse would be close and should handle low frequencies w/o saturating with a few volts swing.

ohhhh this looks doable 10 taps http://www.surplussales.com/item/_ta/155-149496.html
might need the loading on the outputs? headphone amp should drive it

 

[jean-paul]

Impedance is quite low (also when used "in reverse") and they are possiby hard to drive which defies the reason for its use They both are output transformers to drive 8 Ohm loudspeakers and not small signal input transformers. An input transformer could be an elegant but slightly expensive solution though.

With opamps:

http://www.tubecad.com/april_may2001/page2.html

There was a swiss guy in this forum that used a simple non global feedback circuit with opamps and EL84's but I forgot his name.

 

[infinia]

yeah the I know what it is, looking for the turns ratio at a few volts needs big cores! but the HF response might need some extra loading.
IDK seems fun to play with
op-amp with tiny dual winding switcher would lower risk IMO. some of them are very good regarding EMI ripple!

 

[jean-paul]

Infinia, did you check the schematic of the EL84 amplifier ? Just one tube and some resistor values need to be changed in each channel and it is solved.

 

[infinia]

Infinia, did you check the schematic of the EL84 amplifier ? Just one tube change and some resistor vlaues and it is solved.

change the amp gain ? seems too easy

 

[jean-paul]

Simple solutions often work out best