Small unity gain 5V output stage sought....

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Having tested a few audio output ICs (TDA7052A etc), I'm looking for a 'better' solution. (better here being 'better for my application'!)

What I seek is an output stage that will not apply any voltage gain at all (even though th TDA7052 can be set to zero voltage gain, I have the impression that it can't deliver enough grunt quick enough).

I'm using a 5V rail (I'm pretty much tied to this as I'm using another device to host the supply for this circuit - & it only has 5V on it) ...& rail to rail opamps will feed this output stage. The output is a 4 ohm inductive load. The AC signal varies between 1V to about 4.8V ....the frequency range is about 80Hz to 4kHz. I need it to have very little distortion (the output won't be heard, but nevertheless, it needs to be free of sistortion becuase it's causing a string to resonate & for that to happen well, it has to be an accurate representation of the input.

Would discreet components (vs an integrated output IC) be a better solution?

Can someone give me some pointers please...I'm running out of ideas (not that I had many in the first place!!)
 
1Vac requires +-1.4Vp=2.8Vpp at the output.
4.8Vac will require ~14Vpp.
To give this output you will need at least 15Vdc either as a single supply or as a +-7.5Vdc dual polarity supply.
5V has no hope of giving more than 1.5Vac as an output and that into 4ohms or even 4r0 is a mighty ask.
 
Any audio amp IC capable of 4 ohms running off a 5V rail will give almost no output swing. Even a diamond buffer would only give 3.5V p-p, which may be enough. If you want to get really close to the rail, you need complemetary FETs in common source - and that's not something you sit down and 'design' in 5 minutes (an afternoon, maybe).

If it's being used to excite a string, distortion doesn't matter much. It will respond mostly to the fundamental.

What is the app, some sort of guitar sustain gizmo?
 
Thanks guys..perhaps I should have outlined, those were peak to peak figures not RMS!

Yes, it's to exctie a string, but the distortion does matter (if you listen to the output from the guitar, whilst exciting the string with a sine vs a square, there's a massive difference in tone ...a sine sounds pure with no coloration of the string's sound, whereas driving the string with a square isn't a nice sound at all). Same when using the guitar string as the source....if the circuit clips, you hear it in the guitar's output.

Perhaps we're at cross purposes, so let me put my dilemna another way.....

(these are all sine wave, peak to peak figures!)

Having used a sig gen to 'excite the top E string' sufficiently (for non-guitarists - this being the thinnest string with the least amoiunt of ferrous material for the driver to 'get a grip on' - therefore the trickiest to resonate with the driver coil), it needs 5V peak to peak (1.7V RMS) across the 4 ohm driver coil that I've wound (as measured on a scope).

Now to give the preceding AGC circuit enough dynamic range to work with, I'd have liked to have simply had a unity gain output circuit ...ie 1V into the output stage, 1V seen across the coil (again peak to peak figures) 4.8V into the output stage ....4.8V seen across the coil.

I can work with small signals (eg 2.5V into the output stage with a voltage gain of two....but what I'm finding is that the convenient audio amp IC, are a little sluggish at delivering the necessary drive into the string.
 
I can work with small signals (eg 2.5V into the output stage with a voltage gain of two....but what I'm finding is that the convenient audio amp IC, are a little sluggish at delivering the necessary drive into the string.

Audio amp chips lose 4 to 6 volts of signal swing. Operated from only a 5 volt supply you get 1V p-p if you're lucky. If you're unlucky, it won't bias up at all.

I would be looking at discretes, and not be afraid to use a transformer.
 
theAnonymous1 - that's a great 'lead' on the TPA0211 ...it might just be what I'm after (I've spent quite a bit of time on Digikey's site using their filtering mechanism ...but that one slipped me by!)

jcx...unless I'm missing something, it's not obvious to me how the AD8937 work for my needs? I'm looking for something that can deliver circa 1W into 4 Ohms @330Hz.

I've managed to coax the TDA7053A into a form of what I need, but I'd like more grunt in reserve.

here's a quick youtube video I made of the end result (just to give you an idea of what this is all about)....

http://www.youtube.com/watch?v=DXitvBFwgKY&feature=channel_page

5 strings sustain fine...the top E string is a little bit more stroppy! (that said, the circuit was only running at 4.5V vs the 5V I was using on the bench)
 
You can also go discreet: here is an example of moderate complexity and performance. It gives 4.4Vpp output at 0.5% distortion and is unity gain.
The quiescent current is 20 to 30mA and is thermally compensated by attaching D1 or D2 to the output transistors. The other one compensates for Q3.
 

Attachments

  • BuffVIloVo.jpg
    BuffVIloVo.jpg
    147.6 KB · Views: 452
sorry, missed the current requirement - although paralleling 2 packages worth of the ad8397 duals would get the I out it's not real convenient – and the internal compensation is a little light for low gains – needs good bypassing, pwr/gnd planes

another TI chip that looks OK is the OPA569 - the soic-w 20 package is considerably more hackable than the TPA0211 msop
http://focus.ti.com/lit/ds/symlink/opa569.pdf
you can "dead-bug" the soic-20 by bending its legs and then gluing a heat sink to the belly up power pad – I’ve used this upside down mounting with TPA6120

you do need a little more than your supply V if you need 1 Wrms into 4 Ohms so bridged output chips would be good - also reduces rail-to-rail requirement

if you don't have a power limit then relatively simple discrete options become more viable - Class A bias is usually much easier to get low distortion from
 
Last edited:
Sorry for the delay in replying.

Bridged is fine (probably preferable actually) - why? Have you got another contender?

jcx - thanks for those two suggestions. I have to admit I'm drawn to the smaller case type packages as ultimately space is tight here - it's worth pointing out that the final PCB will need to go inside my guitar!
 
Last edited:
....I have to admit I'm drawn to the smaller case type packages as ultimately space is tight here - it's worth pointing out that the final PCB will need to go inside my guitar!

If its small packages your looking for, they are out there. How about SOT 963?:magnify: They come in duel pairs and complementary. :) Obviously Pd will be the limiting factor for any device package chosen for this application.
 
Last edited:
If its small packages your looking for, they are out there. How about SOT 963?:magnify: They come in duel pairs and complementary. :) Obviously Pd will be the limiting factor for any device package chosen for this application.

If I'm reading that datasheet correct...the overall package is just 1mm long
- that's insane! I should set out my stall..... this isn't a commercial ventrure, I'm just a hobbyist with no access to specialist kit (though ultimately I would like to make a PCB of the final using the larger SMT devices).

My TPA0211 arrived today ...oh my, it's incredibly small too (3mm long and that's with 4 pins either side). I think I have an MSOP to DIL 8 adapter somewhere...but it does beg the question...how the hell does this thing dissipate 2W...ok via a heatsink I hear you say...but how do you get a heat sink on something that's just 3mm x 1mm ....ok, so you probably use a heat disappating plane on the PCB.....but this is just me & breadboard at the moment!

I think I'd better go back to the drwaing board...the smallest I feel comfortable working with at the moment is something like an SOIC8 package!
 
I've just remembered, I have three of these left...

http://www.onsemi.com/pub_link/Collateral/NCS2211.PDF

on the face of it, they ought to be perfect. The datasheet says it's capable of delivering 1.5W into 4 ohms at 5V - so what's my problem?

Well I got 5 free samples....when they arrived, I soldered on some wire to allow me to plug this thing into a breadboard - I tried two, neither worked (ie dead as a doornail). So I put the remaining three to one side & moved on . I now figure that their 'dead from the off' state might be possibly to due to me hot gluing my solder joints at the IC's pins (to secure the relatively weak solder connections) ...that might have taken the IC above its maximum sustainable temperature (150C according to the datasheet)

I have two questions to pose....

1. I touched upon this in my last post, but I really don't know how this thing can dissapate 1.5W - that seems like an awful lot of power for something 5mm long & about 4mm wide! How would this normally be done in pratice so the thing doesn't fry (it has inbuilt thermal shitdown protection, but nevertheless 1.5W!!!)

2. If the maximum temperature is 150C ...how on earth are these things solder flowed to the PCB using ovens?


I'll have another go with one of my remaining three left tonight but without the hot glue!
 
Last edited:
also if you're in control of the coil design you can change impedance to match supply V by keeping toal Cu cross section constant but changing wire size/turns ratio

maybe some of the Class D chips would work too - but thes really are'nt generally available hand prototype-able pkgs
 
Last edited:
Well, I got those other three NCS2211s to work - for a while at least (I guess it was the hot glue that killed the other two!) ....this NCS2111 is*almost* perfect.

However, I'm seeing some distortion that for the life of me I can't figure out why. (& any distortion on the signal starts to introduce unwanted harmonics ...as heard on the sustained string)

When I input a sine wave from my sig gen directly into the NCS2211 ...I get a pure sine wave as seen on a scope into the coil (like no distortion whatsoever). This ought to prove that the NCS2211 is fine (not the cause of the problem)

But as soon as I feed it a signal from the sig gen via an OPA2340 (configured as a non inverting opamp with a voltage gain of about 8) I get a small kink in the upper leading edge of the sine wave (as observed anywhere in the signal chain after the output from the OPA2340 (the sine wave is still pure going into the OPA2340 ...ie out of the sig gen). This isn't clipping (as there's plenty of headroom left before it clips)...it's seen at all signal levels (ie if I adjust the output of the sig gen up/ down the kink stays.


Anyone any ideas what the cause of the kink might be?


FWIW, I'm now out of NCS2111s....of the 3 left, one pin snapped of one of them, & the other two I blew (don't know how, but now no workee & they get very hot!)

As for your other ideas....they're out of my comfort zone! (or perhaps I should say 'knowledge zone') so any more meat you can put on the bone will be most welcome!
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.