Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › Some LME49600 implementations
New Posts  All Forums:Forum Nav:

Some LME49600 implementations - Page 18

post #256 of 291
Quote:
Originally Posted by Avro_Arrow View Post

this is the audio jack I usually use, I'm pretty sure it will fit.

http://www.digikey.ca/product-search/en?lang=en&site=ca&KeyWords=sj1-3543&x=0&y=0

 

 

The 5 pin version might be a little better for the input jack, with the internal switch for disconnect grounding:

 

http://www.digikey.com/scripts/DkSearch/dksus.dll?WT.z_header=search_go&lang=en&keywords=%09CP1-3545-ND&x=0&y=0&cur=USD

 

I'm using your Digikey BOM - extremely helpful! - and I noticed that was a 3 pin jack.  Having that input jack grounded when nothing is plugged in has worked out really well on the O2, using the internal switch.  I was going to suggest going with a 5 pin if there was another board re-spin, but on the bigger schematic I see it is already set up for a 5 pin input jack with grounding.  Best I can see on the layout it looks like there are 5 holes.  Needs to be confirmed, though. smile.gif  On the schematic pin 1 of the input jack is a signal line and pin 2 is the grounded sleeve, but both the 3543 and 3545 jacks have those pins flipped, so it might be necessary to do trace cut and swaps on pins 1 and 2 to use these jacks.

 

wakibaki: thanks for the higher res print!  Works great.


Edited by agdr - 9/23/12 at 2:48pm
post #257 of 291
Thread Starter 

This looks to be the right socket if you can't identify one from a regular supplier:

 

http://www.ebay.com/itm/5pcs-1-8-3-5mm-Stereo-Socket-Headphones-Jack-PCB-Panel-Mount-Connectors-2411-/320960009847?_trksid=p2045573.m2042&_trkparms=aid%3D111000%26algo%3DREC.CURRENT%26ao%3D1%26asc%3D27%26meid%3D2258704038311425622%26pid%3D100033%26prg%3D1011%26rk%3D2%26sd%3D221023868653%26

 

Obviously I can't be certain without having one in my hand, but there are numerous other 5-pin 3.5mm sockets on ebay, all with a chrome ring round the hole, unfortunately they protrude too far from the front of the board to permit it to fit in the enclosure, so be warned. I may change the footprint at the next iteration, but that doesn't help here and now.

 

Failing any other source, I do have a limited number of the correct sockets in hand, but I think you will find it cheaper to buy 5 from that ebay source than pay the postage from here, I get away with sending the boards letter mail (£2) because they are flat, but package postage is twice that.

 

w

post #258 of 291

On the board as it is now, signal is on the outside pair and the rest connect to the ground plane.

 

Thanks for pointing that jack out...I will see about incorporating it into my future projects.

I'm glad you found my BOM helpful.

 

Quote:

Originally Posted by agdr View Post

 

The 5 pin version might be a little better for the input jack, with the internal switch for disconnect grounding:

 

http://www.digikey.com/scripts/DkSearch/dksus.dll?WT.z_header=search_go&lang=en&keywords=%09CP1-3545-ND&x=0&y=0&cur=USD

 

I'm using your Digikey BOM - extremely helpful! - and I noticed that was a 3 pin jack.  Having that input jack grounded when nothing is plugged in has worked out really well on the O2, using the internal switch.  I was going to suggest going with a 5 pin if there was another board re-spin, but on the bigger schematic I see it is already set up for a 5 pin input jack with grounding.  Best I can see on the layout it looks like there are 5 holes.  Needs to be confirmed, though. smile.gif  On the schematic pin 1 of the input jack is a signal line and pin 2 is the grounded sleeve, but both the 3543 and 3545 jacks have those pins flipped, so it might be necessary to do trace cut and swaps on pins 1 and 2 to use these jacks.

 

wakibaki: thanks for the higher res print!  Works great.

post #259 of 291

That eBay source for the jack is probably is the best way to go!  I'll get some on order.
 

I'm exploring building mine up as near unity gain (just a current buffer), with +/-6Vdc rails.  I worked up LT Spice sims of the power supply and amp/servo section this afternoon at +/-6vdc all all seems to work very well.  Especially the servo!  I tried adding a +50mV offset to the input signal, then again with a +150mV and -150mV DC offsets.  The output of the sim corrected well, to within +/-0.5mV or so.  Will be very interesting to see how the actual circuit matches up.  There is no sim model for the LME49600 and LME49990, so I'm using the models for the BUF634 and LME49720 instead.  I have a LT1007 in for the servo amp. 

 

I'll try posting the circuit here, but will probably have that same issue of being too small to read.  To get (near) unity gain I'm using the values in the Wire amp, 10K to ground with the 1K feedback resistor.

 

I've wanted to try a reduced voltage version with both the O2 and Wire but never did.   All the chips involved here are specified at 5Vdc, according to the data sheets.  The performance will likely be slightly less on the smaller rail voltages, according to the sheets, but so will the dissipatiion.  I might be able to get away with LM317L and LM337L (the TO-92 version) regulators.  I'm doing the sims with a 12Vac source (transformer) with 5R series resistance.  My big reason for the unity gain version is my particular headphones don't need a lot of voltage gain.  The AKG K550s are full volume at just 80mV rms (114dB/V sensitivity) and the Shure SRH940 at about 1.8V rms.  My sources put out 2V rms.

 

In the plots, green is input and blue output.  The first is 200mV input at 1.2kHz into a 32R load with 0Vdc input offset.  The second is the same thing with +150mV input DC offset.  The third is an ac plot from 5Hz to 500kHz.  The 100K resistor to ground before the input of the LME49600 is a sim model issue.  I found out from sims of the Wire amp the LME49600 model will not converge without some small DC path to ground.  The sim model predicts clipping at around +/-3.9Vdc with a 44R load, with the +/-6Vdc rails.  Will be quite interesting to see if real measurements are close, or if the sim turns out to be off by a mile!

 

 

 

 

 

 


Edited by agdr - 9/23/12 at 9:10pm
post #260 of 291

One more interesting sim plot that seems to show the importance of the relay in addition to the servo.  This is the step response, with the 150mV input DC offset occurring at 1.0 seconds.  The servo settles after about 34 seconds.  The relay would kick in until the servo gets the DC back under control.

 

 

ould kick in until the servo gets the DC back under control.

post #261 of 291
Thread Starter 

Ah, I need to run some sims or tests on the servo/relay circuit myself. When does the relay actually drop out in response to a step input? I don't think it does until the DC offset at the servo output gets quite large (quite a few seconds), since the take-off point feeding the comparators is at the servo output, at the midpoint of the filter intended to keep switching spikes from the chopper-stabilized servo opamp out of the audio.

 

This is not the only function of the relay, it prevents thumps at switch-on and switch-off. Obviously you can speed the whole thing up, but only at the expense of increasing the impact on the FR of the amplifier as a whole. Switching off the output comparatively late is better than never switching it off. This way is better than a (another) simple low-pass filter at the output. Output offset protection is always problematic, there's always a compromise between response time and false triggering due to low frequency program content, I always intended to return to the issue and that of the servo authority at some point, but I've allowed myself to be distracted.

 

post #262 of 291

I haven't tried simulating the relay circuit yet.  I'll give that a shot this weekend.  Looks like a really good implementation.

 

Since all the chips involved seem to go down to +/-2.5V rails from their data sheet graphs (LME49720/10, LME49990, LME49600, LTC1150, LT1007, OPA277) it should actually be possible to use the amp directly off two lithium batteries at +/-3.5Vdc -> +/- 4.2Vdc!  Going with the "transportable" model rather than "portable", and given the quiescent draw, those would probably have to be (larger sized)18650 or 26650s in an external battery holder.  I have both here I use in high power LED flashlights.  The 26650's are about the size of "C" cells at 3900mAhr:


http://www.opticsplanet.com/4sevens-rechargeable-lithium-ion-battery-fs-26650.html?utm_source=google&utm_medium=cpc&utm_campaign=plusbox-beta&gclid=CKDtlfHd0LICFXCmPAodeyUAeg

 

Assuming a 200mA max draw from the amp, that would be 3900 / 200 = 19.5 hours! regular_smile%20.gif  Then assuming "music power" draw vs. sine waves would extend that at least 2x, and probably 3x, the amp might make it 2 full 24hr days on one set of 26650s.  The 18650's go up to around 3100mAhr these days


http://www.lighthound.com/AW-18650-Protected-3100-mAh-Rechargeable-Lithium-Battery-_p_3954.html

 

so would go nearly as long at a smaller size.  With batteries the regulators may not be needed, saving some more voltage overhead, or maybe a pair of 3.3V LDO regulators could be used.  I wouldn't be brave enough to design a charger for lithiums, but they could be charged externally on a commercial wall charger if the battery pack was separate and plugged into an amp somehow.


Edited by agdr - 9/25/12 at 5:48am
post #263 of 291
Thread Starter 

I have this:- http://www.head-fi.org/t/586042/what-i-did-next-with-texas-headamp-chip-a-digital-control-desktop-portable PGA3120/TPA6120 design running from 6 * 16340/CR123A lithiums, but I didn't incorporate the constant current/constant voltage charger in the 49600 amp. I use protected cells in combination with the charger, it requires accurate setting of 2 trimpots, but thus far I've had no problems.

 

I decided to go with the charging scheme after reading a few datasheets and some articles about lithium battery charging and standby use. The batteries are continuously in circuit even when the wall supply is plugged in.

 

You'd have to drill an extra hole in the backplate and bring out a wandering lead if you wanted to plug an external battery pack into the 49600 amp, but you'd have to think hard about how you wired it in since there's currently separate regulation for each channel

 

w

post #264 of 291

Hey that is an interesting amp!  Those 16340s are small enough to get a few in like that, to get the voltage levels up.
 

post #265 of 291

Okay, I'm joining you guys in building this amp. Naturally I will have a couple of questions, so let's see...

 

Why was the LTC1150CS8 chosen? I see it's a zero drift chopper amplifier with low offset and +/-16V supply capability. Are there any other requirements for an opamp in this position? Any alternatives, like would the ADA4638 do as well?

 

That'd be it for now, but there will be more to come.

post #266 of 291
Quote:
Originally Posted by Rodeodave View Post

Okay, I'm joining you guys in building this amp. Naturally I will have a couple of questions, so let's see...

 

Why was the LTC1150CS8 chosen? I see it's a zero drift chopper amplifier with low offset and +/-16V supply capability. Are there any other requirements for an opamp in this position? Any alternatives, like would the ADA4638 do as well?

 

That'd be it for now, but there will be more to come.


I used an OPA211 in mine and it worked fine.

It settles  to 0.1mV after about three minutes.

post #267 of 291
Thread Starter 

The DC servo integrator offset is critical in determining the overall amplifier output offset, so the offset voltage of the opamp used is critical. Since the integrator uses quite high values of resistance to keep the size of capacitors used manageable, the input offset current has quite a big effect on the output offset and it's necessary to calculate its effect when choosing an opamp. 

 

It's a close call in performance terms. The two chips are very close in terms of offset voltage and current, and the ADA4638 is cheaper by a good margin (less than half the price). I'd probably use it in future over the Linear part, particularly as it doesn't require additional filtration at the output, although the filter in the amplifier as drawn serves a dual purpose, and should be left as-is if the ADA4638 is used. Other than that, they're so close that in testing either chip might win, depending on the individual examples.

 

I has a pretty good search round for low offset parts at the time I drew up the design, and ran a spreadsheet to calculate overall offsets, but I didn't see the ADA4638 at the time. Low offset and +/- 16V with a good output swing are the principal requirements, so if you find anything exceeding the performance of the ADA4638 then please let me know.

 

w

 

Of course the actual output offset may in practice be perfectly satisfactory with a range of different parts, and in fact it has been argued that the LME49600 amplifiers with no DC servo are entirely adequate. The amount of offset tolerable is open to question, but nobody would really dispute that a low offset is preferable in principle.


Edited by wakibaki - 3/23/13 at 10:34am
post #268 of 291
Thread Starter 

I believe the Mouser BOM which Avro_Arrow posted in the thread is accurate, but I prefer 565-1577-ND for the electrolytics as they are 15mm tall. This will ensure that they clear the extrusions in the recommended case. I will move the footprint slightly in a future iteration so that taller caps will also fit.

 

This is now a well-proven design with some desirable features.

 

1) It fits an attractive off-the-shelf enclosure and is completely self contained (PSU is on the board).

 

2) It runs off a readily available wallwart.

 

3) It has a switch-on and -off relay to protect headphones from any transients (thumps or clicks) and any excessive (beyond correction) input DC offset.

 

4) It is fully DC coupled throughout with a DC servo and vanishingly small output offset (<0.1mV).

 

5) It employs active parts enjoying the highest reputation for quality and can be built with the very best passives in the signal path.

 

No claims are made for the performance of the amplifier beyond the RMAA and oscilloscope results shown, but if they are to be believed (and I think there is little reason to seriously question them) the noise, distortion and crosstalk are all well below the threshold of audibility, the FR is flat, the output impedance is extremely low (<1R) and the output voltage swing is sufficient to drive all but the most demanding of 'phones, and there is even the possibility of building for higher power rails.

 

There are 3 boards remaining, although I will probably order more when they are gone. Price (non-profit) including worldwide shipping is £6.50 (UKP).

 

w

post #269 of 291

Hi Wakibaki

 

Congrats - great work on this. You have PM.

 

chalk

post #270 of 291

Thanks for the board Fred. I'm looking forward to the build. I don't have any SMD parts in my parts bin, so I'm checking out Farnell. But before I order, does anyone have spare SMD components that would like to sell?  

 

Chalk

New Posts  All Forums:Forum Nav:
  Return Home
Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › Some LME49600 implementations