LM6171 HeadAmp with input buffer (with pics)

[bpribadi 0]

My main purpose building this test board is to test the LM6171 Op-Amp. Some people said it sounds very good, probably the best op-amp in the world. This Op-Amp has a few great features:

 

1. Very high slew rate : 3600 V/uS

2. High output current : 100 mA

 

It is known that slew rate is one of the most important factor of a good amplifier. The higher the slew rate, the better the amp.

 

Just for comparison of slew rate:

 

OPA2134 : 20 V/uS

OPA637 : 135 V/uS

AD8610 : 60 V/uS

 

And for the price of SGD 5.6 per IC (single op-amp), LM6171 is really a temptation for DIYers.

 

But the problem here is, LM6171 is non JFET input op-amp. Means we have to deal with output DC voltage offset. Ussually the approach is to put capacitor at the output to block the DC offset, but we know it will degrade the sound quality.

 

To match the positive and negative input impedance is the better approach, but practically no source (line out) will be able to drive it. Because it means the source has to be powerfull enough to drive a few hundreds ohm or less than a hundred ohm load. Normally line out can only drive 22 kohm or higher.

 

Another problem with highspeed op-amp is oscillation. We have to keep the wiring short to avoid paracitic capacitance and inductance. And to avoid oscillation, we have to use a low value (1 kohm or less) for the feedback resistor. 510 ohm is the recommended value from the LM6171 datasheet. For gain 6.6 as what I have set, I have to use the combination of 510 and 91 ohm. The Thevenin value of those resistors is 77.2 ohm. It means that we have to put 77.2 ohm impedance at the + input. There is no line out would be able to drive 77.2 ohm impedance.

 

So I put a buffer at the input to drive it. The volume pot is put before the buffer. With this configuration, the + and - inputs of LM6171 will 'see' constant impedance all the time.

 

Simple approach, but so far I've never seen anybody took this approach with LM6171. I know that the buffer slew rate (2000 V/uS) is less than the LM6171 slew rate, but to me it is already high enough compare to other design with different op-amp. I think even source the line out will not have such a high slew rate, correct me if I'm wrong

 

 

P1 is to fine tune the + input impedance. Ri1 is for buffer output protection, to avoid short circuit if P1 accidently turned to 0 ohm.

 

I use wall power to get high supply voltage, as we know the higher the supply voltage, the higher the op-amp slew rate. So I will set it to +/- 16 volts. The power supply is the standard LM317 and LM337 combination.

 

So here it is, my test board for the LM6171. Not 100% done, I will still need to connect all the cables, and put it in a nice metal casing (3x6x2 inches). Hopefully I would be able to power it up next week. I will post my impression of how it sounds, or does my approach work to deal with the LM6171 DC output offset.

 

Additional note, I also put 2 jumpers, J1 and J2. They are to test the class A biasing using JFET cascode method. If J1 and J2 jumpered, the LM6171 will be biased to class A operation (like in the PPA). So I'm curios to hear what's the different it will make with LM6171.

 

I just can't wait to test it

 

Board layout:

 

 

 

Edited August 8, 2004 by bpribadi

[Q00 0]

Very neat! Impressive to say the least.

 

Just a quick question, with the transformer being so close and all, wont it affect the circuit in some way?

Edited August 8, 2004 by Q00

[Rameish 0]

You might wanna look into Jan Meier's design (or have you? ;-) The LM 6171 is his choice of op-amp). His current crop of home amplifiers also uses the 634 buffers which if you use the T version will enable you to mount a headsink quiet easily.

 

You might wanna try a 10K volume post as well (as per Meier's design).

 

Here are some links:

 

www.meier-audio.de

 

www.headwize.com

 

the later has an enormous library of diy designs including Jan's

[bpribadi 0]

Very neat! Impressive to say the least.

Just a quick question, with the transformer being so close and all, wont it affect the circuit in some way?

 

Thanks! Well, it will in theory, but I hope it won't be audible, at least. Well this is test board for prove of concept, so I don't know what will be the result I just hope it will sound good, if not, then it will be modified.

 

 

 

You might wanna look into Jan Meier's design (or have you? ;-) The LM 6171 is his choice of op-amp). His current crop of home amplifiers also uses the 634 buffers which if you use the T version will enable you to mount a headsink quiet easily.

 

I haven't look at Jan's design, thanks for the link, it will sure give some additional knowledge Anyway, what is T version? Different packaging of the buffer is it? I will look at the datasheet.

[digi01 0]

great work bpribadi,and great drawing too

 

 

 

[heady 0]

bpribadi, wow! Really cool. Can't wait for you to post your listening opinion.

[bpribadi 0]

Thanks Digi and Heady!

 

I almost blow my BUF634 Somehow I thought BUF634 and HA3-5002 share the same pin configuration, and I use the HA3-5002 pin config to build my test board. I don't have HA3-5002 at hand, only a couple of BUF634. I almost plug in the BUF634 and power them on... lucky I did a final check and I just realize BUF634 has different pin configuration.

 

The board are ready, all cable has been connected, but no casing at the moment.

 

Just note for the audio cable, I found S-VHS cable at SLT (SGD 2.5 / meter), and I like them. They come in a pair of coaxial cables, covered, so looks like a single coaxial. The material is Pure OFC, and have direction on the cable outer insulator. The size is just nice, not too small and not to big. The shield has been proven to be effectively suppres hum. Previously I use normal audio cable (about the same price) for interconnect, and my CMoy suffer from hum, not at home, but in office. After using the S-VHS cable, the hum gone.

 

I'm going to get HA3-5002 tomorrow from Farnell. Since it is much cheaper to buy 10 pcs, I'm going to buy 10, the rest are for my PPA project in the near future. But if any of you need HA3-5002, you can share the purchase with me, since it will be cheaper to buy 10 pcs or more. Just PM me.

 

The power supply is working fine, adjusted to +/- 16.000 Vdc

 

So waiting to get the buffers

 

Bram

Edited August 9, 2004 by bpribadi

[Rameish 0]

Yeah the T version is not a DIP or a Soic version. It's much easier to work with on DIY projetcs.

[bpribadi 0]

I hate to say that the amp schematic above is not working as expected

 

The problem is not with the LM6171, but with the HA3-5002 buffer. I just realized that the 5002 input is not precisely balanced with the supply voltage, but slightly higher, and from the input there is a very small current flow out, about 1.5 uA.

 

So what's the big deal with this 1.5 uA? I put 1 Mohm resistor (which is to high after I relook at the datasheet) at the buffer input, with 1.5 uA flowing out, it produce 1.5 volt at the input . Same voltage at the buffer output, amplified by the LM6171, producing almost 10 volt DC at the headphone output Lucky I checked before I blow my can

 

Now I'm listing a few ways to solve this problem:

1. Adjust the Vs for the buffer until I can get 0 volt at the input (and output). This means buffer and LM6171 must have seperate voltage regulator.

2. Try BUF634 (modify my board).

 

Any more suggestion?

 

Thanks!

Bram

Edited August 11, 2004 by bpribadi

[bpribadi 0]

Aarrgghhh... those approaches don't work either

 

Anyway, look at the bright side , I learn a few things from my experiments which I can share it with you:

 

1. Buffer cannot be use stand alone in audio circuit if you expecting 0 volt at the input and output without the input signal. Means, to avoid input and output offset, buffer must be put in the feedback loop. BUF634 has more balance input circuit than the HA3-5002, it produces about 1/20 DC offset than HA3-5002. But still has to be put in the loop.

 

2. Adjusting the Vs+ and Vs- does reduce the buffer DC offset. But the adjusment is extreme, so it became not applicable. I can get 0 volt DC offset when I adjust the Vs to +4 and -15 (which is ridiculous). About the same value for both BUF634 and HA3-5002.

 

So, my board will keep becoming a test board I have another good thing to share, I didn't hear any audible hum eventhough the transformer is close to the audio circuit. I thing this is due to the shielded input cable. I keep the shield to the last 5mm from the input pin of the IC. Again the S-VHS cable prove it's ability to suppress hum Try this cable for interconnect, they are good for normal purposes. Probably the high end fans would go for silver cables

 

So what's next... I plan to build MINT with my AD8620 and BUF634, while testing the Low Dropout voltage regulator LM2991 and LM2941 (they seems good, but they are expensive )

Edited August 12, 2004 by bpribadi

[mukegile 0]

Bram....., may be its time to take it slow man... ....

cool the head before another one ....... ...

 

i guess don;t ditch the 6171 ideas yet....

u might want to look at one of jan meier's design..

(like what rameish said....6171 is meier's choice of op-amps)

Edited August 12, 2004 by mukegile

[bpribadi 0]

Thanks H!

 

Jan Meier's included to many feature that I'm not interested with. And capacitor in the audio line which is what I'm trying to avoid. I like amp with less components and as simple as possible. I believe less components with good design will give the best sound.

 

Don't worry, I'm not going to give up on the LM6171 so easily, until I found good way to deal with the DC offset

 

Anyway I have planed to build the PIMETA like (sorry, not MINT as what I wrote above) when I got the AD8620 from AD, so it is just an intermezzo