[f] AMB on Facebook

α24 project introduction and development

alpha24 (preamp/buffer, voltage DAC analog output stage, balanced ↔ unbalanced converter)

α24 project introduction and development

Postby amb » May 22nd, 2014, 1:00 am

News

α24 development and testing is complete. Production PCBs are now available at AMB audio shop.

α24 fully-differential line amplifier

(This post was slightly edited and moved here from the γ3 DAC development thread)

The α24 (alpha24) is being developed as a fully-differential line amplifier suitable for use in a balanced preamp, as well as an analog output stage for voltage-out DACs. There is synergy in the two target applications that a common design can serve well in both. The γ3 DAC being concurrently developed will use the α24.

The design strives to achieve high common-mode rejection ratio (CMRR), low noise, low DC offset, low drift, high stability/accuracy characteristics. In addition, as a DAC output stage, it must perform an effective low-pass filter (LPF) function to remove the DAC's ultrasonic artifacts. Finally, it must also contain a balanced-to-unbalanced converter in order to support unbalanced downstream devices.

As a line stage, you can think of the α24 as a fully-differential version of the α20, except α24 is based on premium IC opamps whereas α20 is fully-discrete. The α20 has enough output power to drive headphones, but the α24 is only intended to drive other high(ish) Z inputs, such as a power amplfier or other sound processing equipment.

"Fully-differential" explained

The topology I chose is a modified instrumentation amplifier. This topology, as the name suggests, possess the very performance characteristics that we desire, and is often used for laboratory-grade measurement/test equipment. The usual instrumentation amplifier has a three-opamp topology: a high-CMRR input buffer/gain stage, followed by a balanced-to-unbalanced converter with optional gain or loss:

Image
The standard instrumentation amplifier provides only an unbalanced output, and does not have a LPF function. More importantly, it does not have an output common mode reference voltage control function. This function is needed in the DAC output stage applcation, because the DAC's output pins have a large (1/2 analog supply voltage) DC offset, and we want to avoid using output coupling capacitors.

Thus, I changed the topology of the second stage of the instrumentation amplifier to be a fully-differential amplifier with an optional 3rd-order LPF. The output from this stage is used to drive the balanced XLR analog output. Then, a third stage resembling the standard instrumentation amplifier's second stage is used to convert from balanced to unbalanced for the RCA analog output.

Here is the current schematic:

Image

One benefit of the α24 topology is that you could connect either input to ground and drive the other input from an unbalanced source, and yet the balanced output will provide a fully-differential signal. In effect, it also performs unbalanced-to-balanced conversion, so that when used in a preamp (and driving a fully-differential power amp), the signal would be differential from the preamp to the speaker terminals. No additional conversion circuitry or transformer is needed. Of course, with a fully-differential source (such as the γ3 which will also have a α24 as its analog output stage), the signal will then be fully-differential from the source to the speaker terminals.

The α24 will be offered one channel per board, so that it could be used in a variety of different applications, such as in multi-channel devices. They may also be powered by separate PSUs in a dual-mono (or multi-mono) setup. A low noise dual-rail power supply such as the σ22 will be ideal for it.

The opamps chosen are the OPA1612 for the first stage, OPA1632 fully-differential opamp for the second stage, and OPA1611 for the third stage. These are all high performance opamps that are specialized for audio, and have great reputation for their good sound. These opamps are in SOIC-8 packages, relatively easy to hand-solder, and are commonly available from the usual distributors.

For use in the γ3, the first stage of the α24 is set to have a differential gain of 2, so that the XLR output will provide a pro-level of 4Vrms at 0dBFS. The second stage will have unity gain, and the third stage has a gain of 0.5, so that the RCA output will be redbook standard of 2Vrms at 0dBFS. In a line stage preamp, the first stage gain can be set higher, perhaps to 4 or more.

The LPF function is implemented in R10, R11, C1, C2, C3 and C4. It provides a 3rd-order filter (18dB/octave) with corner frequency of about 100KHz (compared to a 2nd-order filter at the same frequency in the γ2). Where an LPF is not needed, install a wire jumper in R10 and R11, omit C1 and C4, and install small-value compensation capacitors in C2 and C3 for stability.

Of particular interest is the Vocm pin on the OPA1632. By connecting it to ground, the differential output common mode voltage is now referenced to ground. It shifts the output common mode "zero" reference to ground potential (even when there is a large input common mode DC offset), and thus avoids the use of coupling capacitors. Due to tight-matching of devices within the opamps, and due to balanced effective resistances on all opamp input pins, it doesn't even need a DC servo circuit to maintain very low offsets.

Even though there are only three opamps (technically, there are four, because the OPA1612 is a dual unit), there is a lot of performance and functionality here. To replicate it in discrete components would be quite complex, involving a lot of parts (not to mention a big board). I have an idea about what such a circuit might "look like", and it would also need some rare parts. This is why I am not pursuing a discrete solution at this time.

helzerr likes this post.
User avatar
amb
Site Admin
 
Posts: 10684
Likes: 212 posts
Liked in: 513 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: α24 introduction and development

Postby woggy » May 22nd, 2014, 10:58 am

I don't know where to start...
I have been dreaming of an y3 for a long time now...
I just...
I love you ♥ ;)
woggy
 
Posts: 14
Likes: 0 post
Liked in: 0 post
Joined: October 11th, 2011, 12:55 pm
Location: Staffanstorp
Country: Sweden (se)

Re: α24 introduction and development

Postby LV26 » May 22nd, 2014, 11:33 am

Improve Instrument Amplifier Performance with X2Y Optimized Input Filter
http://www.johansondielectrics.com/tech ... 35CEYlBqR0
Everything should be made as simple as possible, but not simpler. // Albert Einstein
User avatar
LV26
 
Posts: 431
Likes: 55 posts
Liked in: 16 posts
Joined: May 7th, 2014, 1:46 am
Location: UKRAINE
Country: Ukraine (ua)

Re: α24 introduction and development

Postby amb » May 22nd, 2014, 12:02 pm

Interesting, but that X2Y input filter require the construction of a bizarre capacitor.

Bill Whitlock wrote an article about a "bootstrapped RFI input filter" for instrumentation amplifiers in the following article. This technique is cool but requires yet another opamp:
http://www.jensentransformers.com/an/ingenaes.pdf

By the way, I neglected to draw input resistors and capacitors in the above schematic. I will post an update shortly.
User avatar
amb
Site Admin
 
Posts: 10684
Likes: 212 posts
Liked in: 513 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: α24 introduction and development

Postby LV26 » May 22nd, 2014, 1:23 pm

amb wrote:Interesting, but that X2Y input filter require the construction of a bizarre capacitor.


X2Y capacitor is a standard part from many companies.
See http://www.x2y.com/mfgs.htm

Digi-Key and others distro have them on stock.

BTW.
I use http://www.findchip.com for search parts for availability.

P.S.
Some very interesting papers there
http://www.x2y.com/filter.htm
Everything should be made as simple as possible, but not simpler. // Albert Einstein
User avatar
LV26
 
Posts: 431
Likes: 55 posts
Liked in: 16 posts
Joined: May 7th, 2014, 1:46 am
Location: UKRAINE
Country: Ukraine (ua)

Re: α24 introduction and development

Postby amb » May 22nd, 2014, 2:18 pm

SMD only... :(
User avatar
amb
Site Admin
 
Posts: 10684
Likes: 212 posts
Liked in: 513 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: α24 introduction and development

Postby MASantos » May 22nd, 2014, 2:43 pm

After soldering the soic opamps, an smd capacitor shouldn't be too hard.

I was in fact wondering if you would consider adding smd pads for the resistors for the a24 and also the y3.
I know that an smd only pcb in probably of the table as I know your preference for through hole from other projects, but smd makes it possible to have very short signal lines, and there are some very nice low noise smd resistors.
MASantos
 
Posts: 75
Likes: 0 post
Liked in: 1 post
Joined: April 9th, 2010, 6:59 am
Country: Portugal (pt)

Re: α24 introduction and development

Postby amb » May 22nd, 2014, 3:36 pm

The annoyance of tiny passive SMD parts is not so much in the soldering, but in that they are easily misplaced or lost. Just a tiny flick of your finger or tweezer, and such a part could go flying and become buried in your carpet, never to be found again. There are active parts that cannot be found in through-hole form, and in those instances I choose parts with hand-solderability in mind. Going by our experience with the γ1/γ2 projects, if we increase the number of SMD parts, it would substantially reduce the success rate among average/general DIYers.

Laying out a PCB to accept both through-hole and SMD parts is problematic, because the SMD pads would block areas where traces could otherwise go through, and SMD pads must be on the top (or bottom) layers, requiring vias for connectivity to internal layers where a through-hole part's pin itself could make such layer jumps. A few SMD parts here and there is probably not a problem, but a large number of "hybrids" would make it impossible to optimize the layout for either type, and negate any advantage of SMD. In this case, such "advantage" is questionable anyway.

That said, I may have to make an exception for the X2Y capacitor. There is a compelling reason for it.
User avatar
amb
Site Admin
 
Posts: 10684
Likes: 212 posts
Liked in: 513 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Updated α24 schematic diagram

Postby amb » May 22nd, 2014, 3:43 pm

The following is an updated schematic diagram. It incorporates the X2Y input filter capacitor, adds the missing input resistors, the board connectors, rail and bypass capacitors. The part ID numbers have also been changed from the previous version.

Image
User avatar
amb
Site Admin
 
Posts: 10684
Likes: 212 posts
Liked in: 513 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: α24 introduction and development

Postby LV26 » May 22nd, 2014, 11:15 pm

Hello,
What is a purpose R1-R3 and R2-R4 voltage divider?
I think we need add 0.1 uF caps in parallel with C11 and C12.
And 10 uF with 100 nF from each V+ and V- pin to ground.
TI recommendation:
- OPA1611 Datasheet, page 10
- OPA1632 Datasheet, page 7

Interesting, but Wolfson recommend C0G/NP multilayer ceramic SMD parts for filter.
- WM8741 Datasheet Figure 83 (Notes 2.)

I`m vote for SMD parts (where they can gain sound quality/performance)
SMD parts the same lice SOIC or TQFP.
You can use 0805 for easy soldering.
SMD = cheap, available, low parasitic.
Everything should be made as simple as possible, but not simpler. // Albert Einstein
User avatar
LV26
 
Posts: 431
Likes: 55 posts
Liked in: 16 posts
Joined: May 7th, 2014, 1:46 am
Location: UKRAINE
Country: Ukraine (ua)

Next

Return to α24 fully-differential line amplifier

Who is online

Users browsing this forum: No registered users and 0 guests

Quick Links

AMB audio DIY resources
AMB audio shop