• These commercial threads are for private transactions. diyAudio.com provides these forums for the convenience of our members, but makes no warranty nor assumes any responsibility. We do not vet any members, use of this facility is at your own risk. Customers can post any issues in those threads as long as it is done in a civil manner. All diyAudio rules about conduct apply and will be enforced.

Bruno Putzeys Balanced Preamp - Group Buy

Status
Not open for further replies.
Jump to Latest
#261
RGN said:
Is LM4562 best choice for dc removal (U3) ?

In the current Group Buy BOM, U3 has been changed from a TL072 to an LM4562 because of the higher offset voltage of the former.

A better choice would be a device with low offset voltage and low bias current such as the AD8622 which has a maximum input bias/offset current 200pA and a maximum offset voltage of 125uV or the OPA2277 which has a maximum input offset voltage of 100uV and a maximum input bias/offset current of 2nA. The downside is the higher cost of either of these opamps compared with the very cheap TL072.

regards,

Rick
Rick
Click to expand...

Hi Rick,

Your calculation is correct at 25 degrees. But the TL072's offset current because of it's Fet input is very temp dependent and can go up to 10nA over the temp range whereas the offset current of the LM4562 will hardly change.

My preamp has an DC output of 1,4 mV and 0,9 mV for resp left and right channel in the +6dB gain position, corresponding nicely with your calculation.
With the TL072 this was around 30mV on both channels after warming up, a substantial difference, implicating that not only Ioffset but also Voffset change substantially with temp. This was THE reason to kick the TL072 out and replace it with a LM4562.

You are right that other Op-Amps might be even better to be used in this position, but with +/- 1mV DC output at 6dB gain, the LM4562 seems good enough and it has some advantage to restrict the number of different components.
On the whole project the few dollars extra will hardly add up to the total costs of this excellent preamp.
 
Last edited:
#262
OPA192

RGN said:
Is LM4562 best choice for dc removal (U3) ?

In the current Group Buy BOM, U3 has been changed from a TL072 to an LM4562 because of the higher offset voltage of the former. A problem with LM4562 is that it has a typical bias current of 10nA and a maximum of 72nA at room temperature (the bias current of the TL072 is a negligible 100pA). Similarly, the input offset current has a typical value of 10nA and a maximum of 65nA.

If you are unlucky enough to have a device near the maximum value it will still produce a largish offset voltage error due to the bias currents flowing in R19 and R16 for the inverting input and R7 and R22 for the non-inverting input.

For example, if the input offset current is 60nA (ie 60nA difference between the inverting and non-inverting bias currents), then the resultant input offset voltage is (60nA x (R7 + R22))= 60nA x 22kilohms = 1.34mV. Admittedly this is better than the TL072 typical offset voltage of 3mV (6mV max) but not a lot better. If you a more fortunate and have a "typical" device then the input offset voltage error due to input offset current is only 0.22mV.

In Bruno's circuit U3A has a gain of 10, but the following stage only has a gain of 0.1 giving a overall gain of 1 at the output of U2/U7A. Further gain is applied by U2/7B which is variable due to R5A/B. If we assume a gain of say 2 for this stage then the input offset voltage of U3A will appear as 2 times that value at the final output.

Using an LM4562 for U3 will result in a preamp output voltage of typically 0.44mV (0.22mV x 2) or 2.7mV (1.34 x 2) if your device has 60nA of input offset current.

Using an TL072 for U3 gives a typical preamp output offset voltage of 6mV (3mV x 2) and a worse case of 12mV (6mV x 2).

If your maximum gain is only unity then the above output offset voltages will be halved, similarly, should you use a higher gain, the output offset will increase. Whether these offset voltages are a problem or not is debatable - it depends how much dc you can tolerate sending to your main power amplifier.

A better choice would be a device with low offset voltage and low bias current such as the AD8622 which has a maximum input bias/offset current 200pA and a maximum offset voltage of 125uV or the OPA2277 which has a maximum input offset voltage of 100uV and a maximum input bias/offset current of 2nA. The downside is the higher cost of either of these opamps compared with the very cheap TL072.

regards,

Rick
Rick
Click to expand...

OPA192?

LP
Dragan
 
#264
The OPA2192 should produce a very low overall offset voltage (theoretically marginally better than the AD8622 and OPA2277) as it has an excellent input offset voltage across the usable temperature range and has a low input offset current. As Hans pointed out above if the LM4562 seems good enough, the audible difference between the LM4562, OPA2192, AD8622 and OPA2277 may be vanishingly small.

regards,
Rick
 
#265
Preamplifier First Stage CMRR
AndrewT in post #244 commented that "If the input is unbalanced, then the common mode as I understand it, is passed along as interference."

This is correct, the more unbalanced the lower the CMRR of the first stage. Bruno has mitigated the problem by using a shared resistor, R10, to provide a large common mode input impedance and low value series resistors (R1 and R9) . Allowing relatively low tolerance (say 1%) resistors (R1, R9, R6, R12) and capacitors (C1, C2) to provide a theoretical CMRR of >80dB @ 10KHz. The attached plot shows the common mode gain with resistors and capacitors set to their worse case values. Since the differential mode gain is unity the CMRR is simply the inverse of the common mode gain. This plot assumes both LM4562 amplifiers (U1) are identical, the track layout has no stray capacitance and the common mode source impedance is 1 ohm.

But precision resistors are required (0.1% or better) to get just 60dB or so of CMRR in the differential amplifier stage based on U2.


regards,
Rick
 

Attachments

#267
Balanced impedance connections.

RGN said:
Preamplifier First Stage CMRR
AndrewT in post #244 commented that "If the input is unbalanced, then the common mode as I understand it, is passed along as interference."

This is correct, the more unbalanced the lower the CMRR of the first stage. Bruno has mitigated the problem by using a shared resistor, R10, to provide a large common mode input impedance and low value series resistors (R1 and R9) . Allowing relatively low tolerance (say 1%) resistors (R1, R9, R6, R12) and capacitors (C1, C2) to provide a theoretical CMRR of >80dB @ 10KHz. The attached plot shows the common mode gain with resistors and capacitors set to their worse case values. Since the differential mode gain is unity the CMRR is simply the inverse of the common mode gain. This plot assumes both LM4562 amplifiers (U1) are identical, the track layout has no stray capacitance and the common mode source impedance is 1 ohm.

But precision resistors are required (0.1% or better) to get just 60dB or so of CMRR in the differential amplifier stage based on U2.


regards,
Rick
Click to expand...
Require the filtering capacitors to be matched. It's impedances that are balanced. The pdf shows 101p & 99p and the effect on the HF CMRR

C1 should be selected to match C2.
I'd suggest +-1% as the minimum standard.
selection to <1% would be very slightly better.
But I don't know which retailers stock 100pF +-1%
 
#269
OK guys and gals,

After sending through the "precision-based" BOM, I've had a reply from the chaps and the cost has risen an awful lot: it's now £93 per board for the components!

However, over £80 of that cost is for the MILFs alone!

My understanding is that should we be able to substitute the MILFs for something else, then the cost is much more reasonable - around £25-30 per board using the other precision components specified in the BOM.

They've also mentioned that some of the Vishay precision resistors are on an 8-10 week lead time...
 
#270
Xavier, as Hans had pointed out in post #258, only R25 to R28 have to be MELF (Mouser for instance stock them for less than .20£), all others don't.
The other resistors I had suggested a few posts ago should also be readily available.
 
Last edited:
#272
Precision Resistors
0.1% 0805 thin film resistors 25ppm temp. coef. are readily available from Digikey and other suppliers. For example, Susumu RG2012P series.
1kohm as used for R2, R3 etc. Digikey RG20P1.0KBCT-ND $0.403 ea. for 10 off (lower for 100off) and in stock.

10kohm Digikey RG20P10.0KBCT-ND. $0.56 ea. for 10 off and in stock

regards,
Rick
 
#275
AndrewT said:
The PCB manufacturer is in the UK.
Digikey is not UK based.
They charge extra for international deliveries and their carrier charges extra for collecting the taxes due.
Click to expand...

Digikey delivers to EU with NO tax and import duties, free shipping above € 60 IIRC.
Same with Mouser (I think they have a € 65 minimum for free shipping).

I use it all the time.

Jan
 
#277
jan.didden said:
Digikey delivers to EU with NO tax and import duties, free shipping above € 60 IIRC.
Same with Mouser (I think they have a € 65 minimum for free shipping).

I use it all the time.

Jan
Click to expand...
I have used Digikey and Mouser.
Digikey and Mouser are USA based and import duties and Value Added Tax MUST both be assessed for each delivery. They can be exempt in some low value imports. That exemption is determined by the tax laws.

They both service the UK and they both charge extra for international deliveries.
They both use a carrier that charges at the door for collecting the tax charge.
Mouser actually charged their fee even though the tax to be collected was assessed as zero !
 
Last edited:
#278
For what it's worth, I have ordered several times from Mouser and the last order was void of any delivery charge (over £33) and import duties / VAT (I ordered through my VAT-registered business where Mouser verify your number and actually exclude the VAT). 3 days after ordering, the stuff arrives, so I've been very impressed with the service.

No doubt larger PCB companies will be able to benefit from this and probably have negotiated better discounts. For the quantities associated with this GB, I can't foresee any issues.

I'm now modifying the BOM keeping the Melfs just for R25 - R28. For the rest, I'm switching to the Susumu 0.1% precision 0805 SMDs, it should lower the price.

I would also suggest to the PCB guys that if they have a better offering on precision resistors (other than Susumu), then I think it's fine as long as they're 0.1% / 25ppm at most.

I'm finding pretty much all the Susumu stuff in stock at Mouser, so availability shouldn't be a problem.

LOL Bill for the MILF typo! Well, what do you expect, with a group buy, we get a good discount 😛
 
#279
AndrewT said:
I have used Digikey and Mouser.
Digikey and Mouser are USA based and import duties and Value Added Tax MUST both be assessed for each delivery
Click to expand...

Incorrect. Mouser for instance ship through Munich and all duties and taxes are included in the quoted prices. And IIRC the last Digikey order I received actually came from the UK. No taxes, no duties.

But 10 years ago it might have been different.

Jan
 
#280
OK, New BoM attached as a CSV file (rename .txt to .csv) and as zipped xlsx file.

All resistors apart from R25 - R28 have been changed to precision Susumu 0.1% 25ppm (apart from the 2M2 ones which are just 1%).

I've written a draft email to Quick Circuits and will send them the updated BoM which I hope will be the last 😉

So whoever is happy with this and confident it's a good compromise and willing to stick their neck on the line, please say so and I will send the email asking for an updated quote.

🙂
 

Attachments

Status
Not open for further replies.
Top Bottom