| Russ White |
Hey Folks,
I have a few projects that have a lot in common, and they are beginning to run together in terms of concerns. In fact the ideas are modular, and build on each other.
Examples:
Twisted Sibling - A balanced line driver/preamp and SE/BAL converter.
Twisted X Overture(TXO) - A balanced symmetrically bridged power amp.
Twisted X Symphony(TXS) - Same as TXO but with two OPA549 working at much lower gain (the gain is handled more by the THS41xx in this case).
Anyway, I have made a whole lot of progress on this stuff, but it does not belong in any single thread, until now. I have decided to launch his thread to discuss the modular approach to all three projects. More projects are planned and in the works based on the modules contained here.
Using the modules I have designed one could create a power amp project of anywhere from 50W to 300W or more....
Anyway here is an article to get people started. I am very interested in feedback from the community here, and ideas. This document is pretty high level. I don't want to lose people in details just yet. :) But this is the place to talk about it, and expand on things. Any important info will be added as a new revision of the document.
I will probably write more detailed articles later should people be interested.
A Very Twisted Missive
Cheers!
Russ |
|
|
| maxw |
| Me too! Its a great concept and very versatile to boot but I really just want the kits already ;) |
|
|
| lgreen |
People have been waiting for this kind of thing to be done by someone who knows what they are doing, esp. an amp that can be high power, the 300W you mention.
It all looks good to me and I think that there will be much interest. |
|
|
| waltona |
Yep, I'm liking it too. I've started to save my pennies now. Keep up the good work.
Al |
|
|
| maxlorenz |
Russ and Brian,
Impressive work and concept! :up:
Maxw,
| quote: | | ...but I really just want the kits already ;) |
LOL :D
Another "obssessive-compulsive builder", like me...
I really have to take the time and open a thread about that subject.
Regards,
M |
|
|
| Ed Lafontaine |
Is it appropriate to ask if DC blocking can be addressed in a coherent fashion?
For instance: "If your XX pre output contains no DC component you may omit the caps."
Is there a "best" place (or way) to manage DC content?
I really like what you're doing Russ. |
|
|
| Russ White |
| quote: | Originally posted by Ed Lafontaine
Is it appropriate to ask if DC blocking can be addressed in a coherent fashion?
For instance: "If your XX pre output contains no DC component you may omit the caps."
Is there a "best" place (or way) to manage DC content?
I really like what you're doing Russ. |
Thanks Ed!
My first choice to block DC would be at the source which is what we do on the Opus.
If you used Opus --> TXD --> JT --> TXO you would nothing other than DC blocking caps which are already on the Opus DAC.
For another source, if you thought you needed it you could add an external CAP. In most cases this will not be necessary, when it is that caps people want are widely varied, and are probably best handled off the PCB.
Cheers! |
|
|
| MashBill |
| I'm very interested! |
|
|
| BillH |
Russ, count me in.
Do the twisted amps handle low impedance loads? I'd like to build an pair of amps of 60 to 100 watts each to drive ESLs. Will the amp's protection circuit handle an overcurrent condition well?
Thanks to you and Brian for your work so far. |
|
|
| Russ White |
| quote: | Originally posted by BillH
Russ, count me in.
Do the twisted amps handle low impedance loads? I'd like to build an pair of amps of 60 to 100 watts each to drive ESLs. Will the amp's protection circuit handle an overcurrent condition well?
Thanks to you and Brian for your work so far. |
Your welcome. :) Its a lot of fun for me too.
If you run the TXO with 2 output (TPM-LM3886) PCBs in parallel per side (4 total TPMs) then you could easily drive just about any load. And would be able to do 200+ watts.
As for overcurrent protection there is nothing I have designed explicitly for that on the PCBs, as that would limit the design somewhat, but the LM3886 itself does afford some degree of protection.
BTW, PCBs have been ordered... we will keep you all posted.
Cheers!
Russ |
|
|
| juma |
Hi Russ,
I thought that your SuSy schematic with OPA549 (http://www.diyaudio.com/forums/atta...tamp=1156203340) would be nice for a 100W/8 Ohm amp my friend wanted me to help him with so I prototyped it quickly today ("fly by wire" - not even a decent p2p). I only changed it so that OPA549 works as an inverting amp and I used OPA1632 (according to datasheet it has a bit more voltage swing than THS4131).
The result was so good that I thought I have to report it back. It was a quick, couple of hours listening and comparing with two designs based on LM3886 (one single chip NI, the other bridged with OPA1632) and I'm almost sure that this one even surpasses them. The OPA549 amp was powered by regulated (LT1083) +-26 V supply. The next step is to build it properly (my friend insists on PCB although I'm a p2p believer).
Thanks for your inspiring effort.
Here is the schematic I used: |
|
|
| BrianDonegan |
| The OPA is nice, but lacks internal common mode offset control, which is why we are using the THS version. |
|
|
| juma |
Yes Brian, you are right. I didn't give it much thought, because in this circuit DC offset on speakers (and to ground) is less than 5 mV.
Anyway, there are situations where this property of THS4131 is important. |
|
|
| Russ White |
| quote: | Originally posted by juma
Hi Russ,
Thanks for your inspiring effort.
Here is the schematic I used: |
Great job Juma!!
Yes the OPA549 at low gain is a great power stage for the Twisted Driver.
We are actually going to produce PCBs for OPA549 as well as the LM3886.
Inverting the OPA549 is not a bad idea at all.
Cheers!
Russ |
|
|
| juma |
| quote: | Originally posted by Russ White
I would run it at even lower gain then you are. You are not really even coming close to pushing the driver.
|
Hi Russ,
I was thinking like this:
Typical source output = 2V RMS = 2.82 V peak
OPA1632 amplifies it 4.4 times = 12.4 V peak (realistic voltage swing with a bit of a headroom)
OPA549's gain = 3.03 (bandwith cca. 300 KHz)
So, gain per side of bridge = 4.4 x 3.03 = 13,33 V/V (10K / 750 R)
Total gain = 26,66 V/V - pretty much standard value for this power but can be even lower, as you said.
Inverted configuration is reported to have less distortion and sounds better, and I can only agree. |
|
|
| Russ White |
| quote: | Originally posted by juma
Hi Russ,
I was thinking like this:
Typical source output = 2V RMS = 2.82 V peak
OPA1632 amplifies it 4.4 times = 12.4 V peak (realistic voltage swing with a bit of a headroom)
OPA549's gain = 3.03 (bandwith cca. 300 KHz)
So, gain per side of bridge = 4.4 x 3.03 = 13,33 V/V (10K / 750 R)
Total gain = 26,66 V/V - pretty much standard value for this power but can be even lower, as you said.
Inverted configuration is reported to have less distortion and sounds better, and I can only agree. |
Yes I actually thought better of my remark and edited it. I think your gain is fine. :)
Good work! |
|
|
| tschrama |
| How did you manage to make those beautiful 3D images?:cool: |
|
|
| Greg Erskine |
| Looks like Eagle and Eagle 3D to me. ;) |
|
|
| psroczka |
Russ,
You told about paralleling OPA 549, did you mean method from TI Application Note or paralleling two moduls for example TXS ? |
|
|
| BrianDonegan |
We would be paralleling two of our TPM-OPA549 (or TPM-LM3886)modules, each containing its own feedback loop and optional servo. It would really be a bridged/paralleled circuit.
So, each channel, if paralleled, would consist of a TXD module (symmetric driver), with the + output driving two paralleled TPM-OPA549 modules, and the - output driving two paralleled TPM-OPA549 modules, for a total of 4 chips per channel.
The kit will consist of just two per channel, but you can add as many in parallel as you like, and they will be available seperately. In fact, each TPM module could be used as a simple SE amp channel by itself. |
|
|
| Russ White |
| quote: | Originally posted by psroczka
Russ,
You told about paralleling OPA 549, did you mean method from TI Application Note or paralleling two moduls for example TXS ? |
Yes the datasheet gives a simple way to parallel the OPA549.
There are two ways to do it with the TPM-OPA549 PCB. You can do it by paralleling two identical circuits using an output resistor on each.
Or, you can set one module up for unity gain and take the output of the actual "Master" module as the input for the follower still using the output resistor.
The second approach is a bit simpler to wire, but the two circuits will not be as well matched, and it may not sound as good as using two identical circuits. |
|
|
| Russ White |
| quote: | Originally posted by tschrama
How did you manage to make those beautiful 3D images?:cool: |
| quote: | Originally posted by Greg Erskine
Looks like Eagle and Eagle 3D to me. ;) |
Greg is correct. :)
Very handy for visualizing PCBs. |
|
|
| pinkmouse |
| I might need to buy a couple of these. I always rated the OPA 549 as the better chip, and I have loads in my bits box. It'll have to work hard to beat the Symasym though! :) |
|
|
| Russ White |
OK here is a simple 2 TPM-LM3886 X amp. Not merely bridged. :)
Here are some pictures which may help visualize what is possible, next I will add four more TPM-Lm3886 modules in parallel with the existing ones, this will easily give me around 250W of power into 4ohms.
Cheers!
Russ
Front... |
|
|
| Russ White |
| Close up of one channel. Both channels have independent transformers and power supplies. |
|
|
| Russ White |
There is still a lot of case wiring cleanup to do, but I am very pleased with the result so far. < 10mv offset.
Each TPM module has a DC nulling servo to keep the idle current under control and generally keep the amp happy. :) These can easily be omitted all together or replaced with a feedback cap (place on PCB for this) which will make the amp unity gain at DC. For me the servo is staying. :)
It sounds as good as anything I have heard with a national chip in it. :drink:
Cheers!
Russ |
|
|
| Panelhead |
Looking good. I hope to build one of these once boards are available. How much do you estimate this will output into 6 - 8 ohms with +/- 30 volt rails?
I have gathered most parts, picked up a pair of 400 va +/- 20 volt trannies. Not waiting on boards, still cleaning up a couple other items. This is something I hope to build sometime early next year. Is that close to your plan commercial release?
Wish I had left the family at Opry Mills one night and tried to swing by and listen last month.. Did not happen, problem with getting them out of the car in the 20 degree weather.
George |
|
|
| Russ White |
Hi George,
Thanks.:drink:
No problem. Family comes first. :) The invite is always open for you.
I will be making a PCB order here very soon (before the holidays). Brian already has a good jump on the parts.
I hate talking about how my own amplifiers sound, but I can't really find any faults in this one. It is quiet when it should be, and loud when you want it to be. :)
As for output power you could expect something like (of the cuff) 80W into 8ohms from those rails. I am using 600VA 20V secondaries get roughly 34V rails (perfect for when I go parallel). But I will be putting in some 500VA 25V secondaries here soon (better for the 8ohm loads).
Cheers!
Russ |
|
|
| BrianDonegan |
Here's a pic of my prototype TXO under construction (monoblocks).
 |
|
|
| erikdebier |
Hi all,
Very interesting project! But can someone give some comments on how it all compares to the MyRef RevC?
Thanks,
Erik |
|
|
| Russ White |
| quote: | Originally posted by erikdebier
Can someone give some comments on how it all compares to the MyRef RevC?
Thanks,
Erik |
Hi Erik,
Its very hard to make a direct comparison, and I believe Brian and I are the only people to have heard the actual TXO with TXD and TPM modules. But we are certainly not impartial and any review either of us gave would be biased. :) That said, here is my biased review. :)
First it is much more powerful than RevC especially into an 8 ohm load as it swing nearly double the voltage. Also you can parallel output PCBs (TPM-LM3886) to get 250W+ power into 4 ohms easy. So for starters it is a much better amp choice for large power hungry speakers. That said, it also lends itself to sensitive speakers because the noise floor is incredibly low especially with balanced input.
Example, I have been borrowing a pair of Jordan single driver speakers that are very sensitive, with TXO silence is black, with my RevC there is just a hint (ear to he cone) of EMI noise and clicks when things like the fridge turn on. Not enough to be objectionable mind you. The RevC is solid, its just that's the difference between balanced input that rejects common mode noise.
Thats about the only direct comparison I will make, but I can say a few more things about the TXO. One of the benefits of the Symmetric nature of the amp (it is not simply bridged) is that because things like crossover distortion, which is always present in any chipamp, appears to the driver in opposing phase, thus common mode, and canceled out. So the TXO has lower crossover distortion. Also, it has incredible dynamic range because the performance of the amp is dominated by the THS4131 which has a great slew rate. The THS4131 is essentially the "soul" of the amp. The LM3886 is really just follower with gain.
In short, the 55W RevC is an excellent choice if you only plan on using single ended gear. I still listen to it regularly.
The TXO is my favorite, of course, and it has proven itself able to drive a much more varied array of speakers. Its character is different, I would say better, but it is also more expensive to make, and not as simple. You pay the price of complexity for quality. The RevC is unrivaled in terms of sound quality for price. TXO is indeed better, but at the expense of cost and complexity.
I hope that helps. Also Mauro is to be acknowledged and thanked for has invaluable contribution to the forum and to my personal knowledge. I rely on him for input often and he is a reliable friend. He designed an incredible amplifier topology and was very unselfish in sharing it.
Cheers!
Russ |
|
|
| jasonb84 |
Just giving this thread a bump as I would like to hear what the latest is at Twisted Pear. Have the new boards arrived and kit prices finalized?
Russ/Brian; How do you compare the sound of the new driver module as a preamp to the X-BOSOZ. Do you think that you will continue to offer both in the future? Or does one consistently outperform the other?
Regards,
Jason |
|
|
| Russ White |
| quote: | Originally posted by jasonb84
Just giving this thread a bump as I would like to hear what the latest is at Twisted Pear. Have the new boards arrived and kit prices finalized?
Russ/Brian; How do you compare the sound of the new driver module as a preamp to the X-BOSOZ. Do you think that you will continue to offer both in the future? Or does one consistently outperform the other?
Regards,
Jason |
HI Jason,
Here is the latest. After a lot of testing a new power supply was designed and I just finished it today. It was worth the work. :) With that behind me now I will making an order for PCBs Monday. So look for these soon!.
XBOSOZ is a very cool preamp, but its PSRR is not nearly as good as that of the TXD. So with the XBOSOZ and attenuator/pot on input you will always get a bit of a hiss unless you have a very quiet PS. With the attenuator on output XBOSOZ is much better, but also has higher output impedance.
The TXD is much easier to build, and does not require the DC blocking caps that the XBOSOZ does. In a word, it is more practical.
TXD fills an important role in our lineup, that of balanced driver, SE/BAL converter, and the key component to our "Twisted X" amps.
Cheers!
Russ |
|
|
| Russ White |
| Here is the new power supply I just designed under testing. :) So far so good. That's 1.3KVA worth of trafo in there. :D |
|
|
| nvrgdenuf |
Brian or Russ,
Beings I can't seem to get an email response from your website, what is the expectations on delivery of the lpm3886 kits pre-ordered and pre-paid? I was surprised that even though I received acknowledgement from paypal, nothing from your site for payment. I want to make sure it was received.
andy |
|
|
| Russ White |
| quote: | Originally posted by nvrgdenuf
Brian or Russ,
Beings I can't seem to get an email response from your website, what is the expectations on delivery of the lpm3886 kits pre-ordered and pre-paid? I was surprised that even though I received acknowledgement from paypal, nothing from your site for payment. I want to make sure it was received.
andy |
Hi Andy,
I am sure it was received. :) Brian usually sends replies right away, sorry if yours got missed. No worries, everything will be straight.
The PCBs have been ordered and you are right at the top of the list when they get here(usually around 2 weeks).
Cheers!
Russ |
|
|
| nvrgdenuf |
Great!
I am really sort of taking a chance on these for my setup. Not that the amps aren't fantastic, but they will be driving the Hawthorne Audio 15" coaxial Open Baffle in a Home Theater that doubles as a multi channel listening environment. They are sensitive at about 96db. I am hoping to do five channels with the basic amp modules and two Open baffle woofers with the TXO's bridged.
thanks!
andy |
|
|
| Russ White |
| quote: | Originally posted by nvrgdenuf
Gmulti channel listening environment. They are sensitive at about 96db. I am hoping to do five channels with the basic amp modules and two Open baffle woofers with the TXO's bridged.
thanks!
andy |
That sounds very reasonable to me :)
In fact, I am planning on building some active speakers using just the power modules. They should work very well in that sort of duty.
Cheers!
Russ |
|
|
| cjd |
Heh.
So I finally get back working on my project(s) and find you're playing with the "modular" approach I had out there ages ago. Glad someone else sees the sense in it! I'll be running some commercial monoblocks off the X-BOSOZ/JT probably by next weekend, but still want to get my own amp stages running.
I have a passing interest in the amp stage(s) you're outlining here as well as I'm continuing to have difficulty with mine (parts should be here Wednesday for more thorough troubleshooting) - or maybe just your "bridge" portion. Still think I would like to keep the X-BOSOZ up front, JT after that, with monoblocks at my speaker(s). I think the actual chip-amp package(s) I have work properly and it's the bridge I'm having trouble with, so I may be curious to try just that - possible?
Would have pinged you on your support site about this but never got e-mails from the forum software (either a failure to send or a failure at my e-mail host and overzealous spam filtering, the latter being my primary suspect - why it thinks I need a bigger ... and yet don't need my twisted support is beyond me. :confused: :clown: )
C |
|
|
| Panelhead |
Found a small package from Brian in the mail. It held the boards to build a Twisted Driver, power supply, and Power modules.
Guess it is time to print out the BOM and start gathering parts. The modular approach is different. Plus all the nice featues of this.
Hope to see pics soon of these. Mine is waiting on me.
George |
|
|
| BrianDonegan |
Here are my monoblocks. I made them symmetrical (get it? ;)). These are actually using the protoype boards, except for the PS. That's a 400VA 20v+20v Antek trafo. So, waiting on a fuse holder for one of them (the one pictured open). I also used two bridges instead of one, but just cause I have a bunch. the heatsink doesn't get even warm, so probably overkill.
 |
|
|
| Panelhead |
| Where did you pick up the cases? |
|
|
| BrianDonegan |
| Those are from Antek as well (aka Par-Metal). Mine are 12x12x3 (small). |
|
|
| jleaman |
| quote: | Originally posted by BrianDonegan
Those are from Antek as well (aka Par-Metal). Mine are 12x12x3 (small). |
exactly where they are from ( great prices too )
Them champaign, looking chassis look clean, nicely laid out and just super nice :D good job..
Jase |
|
|
| protos |
By the way, I found the choice of motto/sign in greek in your workshop interesting.
ΣΚΕΨΟΥ = THINK
How come? |
|
|
| BrianDonegan |
Ah, I was wondering if anyone would notice that.
I was thinking of having a little contest, this first that could tell me what it is would win, but now you have done the hard bit (translation). |
|
|
| protos |
| So no prizes?:bawling: |
|
|
| jleaman |
Does any one know the price for the Se balanced converters ?
Good work on the chassis Brian, keep up that good work.
J' |
|
|
| BrianDonegan |
Maybe a grab bag o'parts. But I think a simple Google search will yield the answer at this point. When I was thinking of the contest, I suspected you would have an edge over everyone else ;)
| quote: | | Good work on the chassis Brian, keep up that good work. |
Thanks!
| quote: | | Does any one know the price for the Se balanced converters ? |
What SE-Bal converters do you mean? |
|
|
| jleaman |
| quote: | Originally posted by BrianDonegan
What SE-Bal converters do you mean? |
the ones in your amps ? or did i miss read your first post. ? |
|
|
| BrianDonegan |
| Ah, those are the TXD boards, which are the heart of the TXO amps. They make them symmetric, not just balanced. |
|
|
| SheldonD |
Russ:
I have some questions about TXO and TXD.
also waiting for BOM and Manual for TXO.
1. What is the IC for current balancing in 3886 schematic ?
I have a few 3886, 3875, 4780 amps from chipamp, I thought to add the circuit to them.
2. You state that the TXD could be set up with gain of less than one, if so would it not be feasible to do so and then run the 3886 at higher gain for increase stability? After all, it is the final overall gain that is important
3. I was considering to build Aleph p1.7 preamp but with TXD
but it seems to be redundant to do so if I build TXO-2. So a passive preamp with Vol control and input selector seems to be
all that is needed.
All the sources are single ended but having balanced at the 3886
seems very desireable.
I will be building Pearl phono as well with a seperate PS chassis
which will be inthe preamp chassis.
Sheldon |
|
|
| Russ White |
| quote: | Originally posted by SheldonD
Russ:
I have some questions about TXO and TXD.
also waiting for BOM and Manual for TXO.
1. What is the IC for current balancing in 3886 schematic ?
I have a few 3886, 3875, 4780 amps from chipamp, I thought to add the circuit to them.
2. You state that the TXD could be set up with gain of less than one, if so would it not be feasible to do so and then run the 3886 at higher gain for increase stability? After all, it is the final overall gain that is important
3. I was considering to build Aleph p1.7 preamp but with TXD
but it seems to be redundant to do so if I build TXO-2. So a passive preamp with Vol control and input selector seems to be
all that is needed.
All the sources are single ended but having balanced at the 3886
seems very desireable.
I will be building Pearl phono as well with a seperate PS chassis
which will be inthe preamp chassis.
Sheldon |
Hi Sheldon,
Good questions, here are some quick answers.
1) LF411 is used for the servo.
2) Certainly something one could try, but my experience was that best stability was achieved when I increased the bandwidth of the power stage (the LM3886 or similar) but lowering its gain so that the power bandwidth of each opamp (THS4131 and the LM3886) was closer together. This resulted in nice low values for the compensation caps(C7 and C8). This is especially true for parts like the OPA549 which I have used with only 6db gain.
3) Passive attenuation into the TXO is indeed excellent. the input impedance is 100K so a 10K pot or equivalent attenuator is fine. Also the Joshua Tree suits it very well with it constant output impedance of 750R. This is my current favorite scheme Opus --> JT ---> TXO.
Brian and I are working on the documentation, and as always we will be happy to answer questions.
Cheers!
Russ |
|
|
| SheldonD |
Russ:
Thank you for the quick response.
So for ques 2. It is important to balance GW for both ICs.
Does this mean the least distortion as well?
To mee the specs for 3886 seemed overall to be better than
OPA549.
SheldonD |
|
|
| Russ White |
Sheldon,
Your welcome. Glad to do it.
Yes, the faster the power stage can react to the input from the global feedback loop the less distortion you get. :) This is actually one of the things I simulated over and over trying to find a good balance.
The OPA549 is actually an excellent power opamp, but not at high gain. keep it < 4 X and it is really great. Above that and its GBWP goes to ****. OPA549 actually suites this topology very well. The LM3886 is also excellent(and less expensive), and I don't think I would put one over the other. I use the Lm3886 at 10X gain which is the lowest the data sheet recommends.
The TXO as designed is very stable. So much so that neither Brian nor I have ever been able to coax it to oscillate. It also runs very very cool. Cooler then we expected for a bridged design. The servo helps in that regard a lot.
Cheers!
Russ |
|
|
| Russ White |
I meant to say that idle it runs cool. It will of course dissipate heat normally under load based on the load.
Cheers!
Russ |
|
|
| mr.duck |
These regulators, they power the power amps? The heat sinks seem way too small...
How does the balanced opamp as an input buffer improve performance? I imagine it can improve the S/N ratio but does it do more to improve sound quality? Perhaps it colors the sound in a pleasing way? (nothing wrong with that for hi-fi)
edit: i think i just got the point of it, it keeps the whole amp in a balanced (differential?) topology? |
|
|
| Russ White |
| quote: | Originally posted by mr.duck
1)These regulators, they power the power amps? The heat sinks seem way to small...
2) How does the balanced opamp as an input buffer improve performance? I imagine it can improve the S/N ratio but does it do more to improve sound quality? Perhaps it colors the sound in a pleasing way? (nothing wrong with that for hi-fi)
|
1) No, those power the TXD only, not the power stage.
2) The TXD (THS4131) "IS" the amp, not just a buffer. It actually forms the dominant symmetric feedback loop. The LM4562 can be looked at as a sort of buffer, but it is in fact an instrumentation amp. The benefit is extremely low distortion and high dynamic range coupled with very good PSRR. The amp also rejects common mode noise very well.
Cheers!
Russ |
|
|
| BrianDonegan |
| quote: | | These regulators, they power the power amps? The heat sinks seem way too small... |
| quote: | | 1) No, those power the TXD only, not the power stage. |
Incidently, it is possible to use the regulated section for amp power, but you need to mount the FETs under the board and mount the whole thing on a heatsink.
The FETs are STP80NF55-06/STP80PF55 from ST (55V/80A/300W). You can use other FETs for higher current if needed. We chose these for optimum impedance performance on the supply (<0.018ohm Ron). |
|
|
| Pathmark |
Russ & Brian,
Thanks again for great kits, so far 1 orginal 3886 kit and now a TXO kit in standalone config, no TXD, working great. My questions is regarding buffer amp.
I have read many posts regarding the "improvement" with the use of buffer. Unfortunately, my attempts to build NUUK discrete buffer are not so succesful.
So my question is does the TXD fill this role? If yes do i need two, one for each channel?
Thanks,
Pathmark |
|
|
| Russ White |
Hi Pathmark, and thanks.
Do you actually have a TXO? Or a pair of TPMs?
If you actually have a TXO it already has the TXD module included for each channel (it is key), and will not require any other buffer as the LM4562 on the TXD acts like a buffer.
Yes, you certainly could use the TXD as a balanced output buffer/line driver (input could be SE). With single ended output it would work fine but it would not be as optimal.
You could actually easily use a single TXD PCB as a single ended stereo buffer by omitting the THS4131 and placing resistors to alternate pads (I have actually done this).
But one word of advice. I bet the TPM would actually sound just great without a buffer, and perhaps better. The reason is that the input impedance is already quite high, and it does not require much current. So a buffer is kinda wasted. Inverting chip amps or low impedance chip amps are a whole different story. They need the extra current.
Cheers!
Russ |
|
|
| BWRX |
| quote: | Originally posted by BrianDonegan
The FETs are STP80NF55-06/STP80PF55 from ST (55V/80A/300W). You can use other FETs for higher current if needed. We chose these for optimum impedance performance on the supply (<0.018ohm Ron). |
Are these FETs used as the pass element in a linear regulator? |
|
|
| Russ White |
| quote: | Originally posted by BWRX
Are these FETs used as the pass element in a linear regulator? |
Yes this is true. It just one factor in the result. The actual output impedance is still plenty low, but in fact the design of the TXD does not actually require low output impedance.
Cheers!
Russ |
|
|
| BrianDonegan |
| Reading that, it was worded poorly. I did not mean to imply the the Ron was the impedance of the PS. |
|
|
| Pathmark |
Russ,
THanks for reply, yes I have TPM modules (2). So are you saying that in current stand alone config no buffer is required?
Mike |
|
|
| Russ White |
| quote: | Originally posted by Pathmark
Russ,
THanks for reply, yes I have TPM modules (2). So are you saying that in current stand alone config no buffer is required?
Mike |
No problem. No you should not need one at all unless your source has very high output impedance.
You could actually just use a 10K pot in front of the TPMs for volume control.
Cheers!
Russ |
|
|
| BrianDonegan |
I think we just found a problem.
The schematic, as well as the part label, states that R11 should only be used with the TXO. This is not the case. It should be used in all configurations, giving a 100K input impedance.
We are making the needed updates to the docs. Sorry. |
|
|
| Pathmark |
Brian,
Not smart enough to know this would be a problem I have been using w/o R11. I am using a passive volume control (autoformer type) being fed from a DAC.
What problems should have been apparent w/o R11? What changes should I expect w/ R11?
Pathmark |
|
|
| Russ White |
| quote: | Originally posted by Pathmark
Brian,
Not smart enough to know this would be a problem I have been using w/o R11. I am using a passive volume control (autoformer type) being fed from a DAC.
What problems should have been apparent w/o R11? What changes should I expect w/ R11?
Pathmark |
Actually probably none. The input impedance of the amp would just be very high, which may not hurt anything.
Adding the 100K resistor can help reduce line noise if you have long lines to the amp. In your situation you may be best just to leave it out if you are not observing any issues.
Cheers!
Russ |
|
|
| MashBill |
Russ and Brian,
I have have been following this thread with interest, but I too am confused about the various modules you offer. Maybe you could put some block diagrams on your website showing what modules would be needed for a preamp, amplifier, integrated amplifier, standalone DAC, etc.
I have a few other projects in the works right now, but when I finish them I want to build a standalone Opus DAC.
Bill |
|
|
| BWRX |
| quote: | Originally posted by Russ White
The actual output impedance is still plenty low, but in fact the design of the TXD does not actually require low output impedance. |
Most low current draw circuits with high PSRR don't require low output impedance, but it doesn't hurt either.
| quote: | Originally posted by BrianDonegan
I did not mean to imply the the Ron was the impedance of the PS. |
That's all I was getting at :) I couldn't find a schematic of the supply so I had to ask. |
|
|
| BrianDonegan |
| quote: | | Most low current draw circuits with high PSRR don't require low output impedance, but it doesn't hurt either. |
Yup. The TXPS is meant to be a general purpose supply, usable for a variety of applications. High current unregulated, medium and low current regulated, and various combinations of the two. |
|
|
| BrianDonegan |
| quote: | | I have have been following this thread with interest, but I too am confused about the various modules you offer. Maybe you could put some block diagrams on your website showing what modules would be needed for a preamp, amplifier, integrated amplifier, standalone DAC, etc. |
Keep trying to get to this, but there are never enough hours in the day. I will try to get this done this weekend, as it's been bugging me for a while. |
|
|
| SheldonD |
Hi Russ I have further questions re TXD
1. I have 2 sets of TXD boards with which I wish to drive two amplifiers:
a. TXO-2 with LM3886
b. TXO-4 with LM4780
the input for each amp will be single end passive preamp output. I believe the answers you give will apply to both amps.
The amp modules are from Chipamp.com and I will have to add the servos.
2. From your documentation and my previous questions the preferred gain at the amp modules should be 10.
3. the gain as the input TXD will be SE so what should the resistor values be to maintain a "line-level" input/output?. Secondly may it be optimum to rebalance the gains with respect to TXD and the amp modules.
4. As the TXD will be both: a. SE to balanced AND b. preamp
some of the statements regarding C7.C8 , the Jumpers and the gain values set by resistors etc seem contradictory.
I am more than a bit confused.
I am suffering from "paralysis thru analysis"
5. TXD ICs are both run with +/- 10 Volts
Thanks,
SheldonD |
|
|
| Russ White |
Hi Sheldon.
Sound like a couple fun projects.
You seem to be thinking exactly down the right lines.
Here are the important points:
The gain of the TXD sets the gain of the entire amp, and with single ended input(-IN tied to GND) the output will be 6db less than with balanced input(with the level).
So lets use 2Vpp inputs as an example. With balanced input that would be doubled so 4Vpp. 20db gain is 10X voltage gain. 10 X 4vpp is 40Vpp(as a reference)
To get that same gain with an equivalent SE signal you will want to add 6db of gain. That works out to 26db or 20X voltage gain.
Now RF and RG have been chosen along with C7/C8(compensation) for stability. So I would not adjust these, instead its simpler to adjust R1 and R2.
G = ( RF / RG ) * ( 1 + (2R2 / R1) )
Choose R1 and R2 for a factor of 2 (R1= 4.7K and R2= 4.7K should be fine).
1 + (4.7K / 4.7K) = 2
Make sure R1 and R2 resistors are as well matched as possible. At worst 1%.
10V rails is just fine. :)
I hope this helps you get up and going. :)
Please share your results!!!
Cheers!
Russ |
|
|
| Russ White |
Also, now that I look at my own schematic I remember that the recommended values are:
RG=2.2K RF=22.1K
You could easily get the same 6db boost for SE input by halving RG to 1.1K, 1K is certainly fine here too.
That would get you the same ~ 20V voltage gain as adjusting R1/R2 as in my previous post.
I still like my first suggestion best (because the interstage impedance remains unchanged) but either approach should produce excellent results.
For TXO with RF=22K C7 and C8 need to be at least 220pf. You can go up to around 330pf with no audible penalty.
Cheers!
Russ |
|
|
| BrianDonegan |
| And to come full circle, the TXD (as well as the TXD kits within the larger TXO kits) kit comes with RG of 2.21k - 0.1% (standard value - 10X gain) as well as 1.1K - 0.1% (for 20X gain). ;) |
|
|
| tryonziess |
| russ this component setup is great. now, how many of your output lm3886 boards do you think can be strung along before the input devices need tweeking. would the control and power units handle 8 lm3886 chips in a mono configuration without overloading the control and power supplies. that would yield about 400 watts to the speaker. ie, 4x4 parallel then bridge for a total of eight. i currently run two line arrays with 20 3.5 inch drivers in each. as i add power, to a limit, the overall sound is improved dramatically. are boards ready to ship currently. anxiously waiting to get this project in high gear. |
|
|
| BrianDonegan |
| Something like this? |
|
|
| Russ White |
| quote: | Originally posted by tryonziess
russ this component setup is great. now, how many of your output lm3886 boards do you think can be strung along before the input devices need tweeking. would the control and power units handle 8 lm3886 chips in a mono configuration without overloading the control and power supplies. that would yield about 400 watts to the speaker. ie, 4x4 parallel then bridge for a total of eight. i currently run two line arrays with 20 3.5 inch drivers in each. as i add power, to a limit, the overall sound is improved dramatically. are boards ready to ship currently. anxiously waiting to get this project in high gear. |
The TXD could easily drive 8 LM3886s at 100K input impedance each. The TXD should be good down to a load of around 600ohms easy. So you could theoretically drive something like 50-60 of them :) Not that you ever would... :D
You actually should be able to do more than 400W with just three in parallel for each half channel. So 6 LM3886(TPM modules) per channel. Assuming your PS is capable of supply the current. :) The TXSPS would work just fine(one per channel) but you would need to calculate the trafo accordingly.
Cheers!
Russ |
|
|
| tryonziess |
| ok, monday morning i am going to order these toys and have some growup fun. i assume every available configuration is easily spelled out. will 625 va at 25-0-25 work for the 300 plus. i have a few on hand. national does not specify more than 30 volts for multiple parallel or bpa. is there a land line number for asking detailed questions or just email and the forum. thanks |
|
|
| Russ White |
| quote: | Originally posted by tryonziess
ok, monday morning i am going to order these toys and have some growup fun. i assume every available configuration is easily spelled out. will 625 va at 25-0-25 work for the 300 plus. i have a few on hand. national does not specify more than 30 volts for multiple parallel or bpa. is there a land line number for asking detailed questions or just email and the forum. thanks |
That trafo should work just fine. The few extra volts should not hurt anything especially with three in parallel.
You can email me directly if you like. |
|
|
| Javin5 |
Brian or Russ
What is the difference between the stereo version and the dual-mono version of the TXD?
The website shows both as 2 TXD Kits plus 1 LCPS Kit. Should it be 2 LCPS Kits for dual-mono? |
|
|
| Russ White |
| quote: | Originally posted by Javin5
Brian or Russ
What is the difference between the stereo version and the dual-mono version of the TXD?
The website shows both as 2 TXD Kits plus 1 LCPS Kit. Should it be 2 LCPS Kits for dual-mono? |
Actually I think thats an incorrect artifact. The TXPS takes the place of the LCPS in the TXO as it has a built in regulator for the TXD (that was what the LCPS was for).
We will get that straightened out. :)
The difference is that the dual mono has two TXPS and the stereo just one. :)
Cheers!
Russ |
|
|
| nvrgdenuf |
I have the stereo pair with the power supply that I had built for me and haven't even hooked up. Going different direction. Cost$126. If somebody wants them for $100 shipped email me.
drptpcpe@gmail.com |
|
|
| Russ White |
| quote: | Originally posted by nvrgdenuf
I have the stereo pair with the power supply that I had built for me and haven't even hooked up. Going different direction. Cost$126. If somebody wants them for $100 shipped email me.
drptpcpe@gmail.com |
Are you sure thats not a RevC or just a pair of TPMs?
That would be a bargain for even a TXO-2. Heck its a bargain no matter how you look at it.
Cheers!
Russ |
|
|
| nvrgdenuf |
Its the LM3886 stereo amplifier complete kit. (1) TXPS kit and (2) TPM-LM3886 kits.
andy |
|
|
| SheldonD |
| quote: | Originally posted by SheldonD
Hi Russ I have further questions re TXD
1 . TXO-2 with LM3886
the input for each amp will be single end passive preamp output.
SheldonD |
[QUOTE]Originally posted by Russ White
[B]Also, now that I look at my own schematic I remember that the recommended values are:
RG=2.2K RF=22.1K
You could easily get the same 6db boost for SE input by halving RG to 1.1K, 1K is certainly fine here too.
That would get you the same ~ 20V voltage gain as adjusting R1/R2 as in my previous post.
I still like my first suggestion best (because the interstage impedance remains unchanged) but either approach should produce excellent results.
But according to your Doc R1 should be omitted so that the gain can only be set for TXD with Rf and Rg ???
For TXO with RF=22K C7 and C8 need to be at least 220pf. You can go up to around 330pf with no audible penalty.
Could you explain this further what are the considerations regarding C7 and C* with respect to R22? Is larger better?
I have further questions re TXD looking at the TXD prototype pic
there is no R01 and R02 (output resistor) in the BOM you have value of 1K.
On the schematic for TPM you have input resister R6 as 3.3k, as TXO change to 221ohms.
This leads to another concern:
on page 13 of BPA-200 doc by National semiconductori they suggest input to LM3886 should be equal to R2 (TPM schematic)
if using TXO does it matter wht the individual values are for the output and TPM input as long as the total is correct?
Also R10 output from Servo should bre 10x feedback resistor R1 of LM3886.
I will be using LF412 to feed 2 LM3886.
Getting back to gain settings: to keep the proportional gain of the TXC and TPM the same: I propse to mutiply each by root 2=1.414 .
I would like to follow the guidelines of BPA-200 as well as the work you have done.
BTW in TPM BOM C7 is missing as well as the alternate 221 ohm for R6.
Thanks |
|
|
| BrianDonegan |
Digesting your post, but, small bites:
TPM BOM shows a 220pF FKP2 cap for C7
Schematic in the user manual is out of date, we will get it updated ASAP. The schematic on the page is accurate. R6 is always 3.32K, no 221R alternate value. |
|
|
| SheldonD |
Thank you for quick response.
On the new schematic it staes that R1 could be reduced, not by much though as G= (1+R1/R2) as it stands G=(1 +10/1)=11
R1 could be 9k0 to yeld 10
Lm3886 gain per data sheet should be 10 or more.
BTW I had sent some of the same questions to Russ a few days back through the DIY email. There was no answer. I have noticed a few posts about this. There may be something wrong.
Regards |
|
|
| Russ White |
| quote: | Originally posted by SheldonD
Thank you for quick response.
On the new schematic it staes that R1 could be reduced, not by much though as G= (1+R1/R2) as it stands G=(1 +10/1)=11
R1 could be 9k0 to yeld 10
Lm3886 gain per data sheet should be 10 or more.
BTW I had sent some of the same questions to Russ a few days back through the DIY email. There was no answer. I have noticed a few posts about this. There may be something wrong.
Regards |
HI Sheldon, I am sorry for my Tardy answer, I started a reply and some how sent it to drafts instead of sending it on. I was in the midst of finishing up Opus (our new DAC) etc and have been quite busy.
Which schematic are you looking at?
I am not sure why you are computing gain like that. This is like an inverting amp for calculating gain. So 10K RF and 1K RG = 10X voltage gain.
Cheers!
Russ |
|
|
| Russ White |
Look at page three of this document:
http://www.twistedpearaudio.com/docs/txd/txd-doc.pdf
It should answer you questions. If it does not then please post again. :)
Remember that with the TXO the gain of the TXD module sets the gain of the entire amp. The gain for the TPMs should stay at 11 (10K + 1K) as that has produced very good results. The gain of the TPMs only effects the voltage that the TXD has to swing (which will not be much) along with the bandwidth for the power stage, too high a gain and you loose some bandwidth. |
|
|
| Russ White |
| quote: | Originally posted by SheldonD
Thank you for quick response.
R1 could be 9k0 to yeld 10
Lm3886 gain per data sheet should be 10 or more.
Regards |
Oh yes, I think i understand your post better now. That note on the schematic was made prior to me change the feedback resistor to 10K. It had bee 22.1K which also worked fine, but the 10K seems better to me. :)
10 is indeed the minimum gain, but the LM3886 seems quite happy with its little loop at a gain of 11. :) Perfectly stable.
I should have deleted the note.
I just got in from mowing the lawn so it took me a bit to recover from the heat stroke. :)
Cheers!
Russ |
|
|
| Russ White |
The value of C7 and C8(TXD) are computed to give a high pass filter with RF of the TXD. 220pf is as low as I would ever go. 280pf is fine as is 330pf. The higher the value the lower the filter and the more phase shift at higher frequencies.
C6 on the TPM also forms a low pass, but you should never use it with the TXO.
The rule for the servo resistor is that it should be "at least" 10X the FB resistor. It is absolutely fine for it to be more, at least with the op amps I am using.
This is not a BPA200 so while some of the info in that document will apply, a lot of it will not.
If you use the same parts Brian supplies with the kit (adjusted for the gain you want) you will be golden.
I would not touch the TPM, leave it at G=11, just modify the TXD.
Also, while the 3.3K resistor is fine on the non-inverting input but is optional, it does not hurt to have it there, but it can be replaced by a jumper or lower value resistor. The non-inverting input of the LM3886 has a very high impedance. The TPM standalone(not TXO) is more like a regular chipamp, where you want that 3K3 to work with C7 to form a low pass to reject high freq noise etc. In TXO that low pass would cause instability and is omitted.
Let me know if there is more you need.
Cheers!
Russ |
|
|
| SheldonD |
Thanks guys.
Things are starting to clear up.
As I see it to compensate for SE (not balanced) I have a choice to increase gain at TXD and/or TPM.
Going along with your original design you had balanced the gain
across both modules to provide the least distortion. I am inclined to maintain the ratio across both.
So again: I propose to share the gain increase across both TXD and TPM. One advantage is that in general one should have circuits running near the centre of their "happy" range. Not at the boundarieswhere things like change/stolerance on components and line voltage and loads could affect things..
What I am still concerned about is the relative values of resisters as per National Semi BPA-200. Your values are out of line but I
realize the TPM is inot in Isolation but is part of a greater loop
with ths4131.
I also realize you both have a successful build but balanced.
One thought is for me to build a SE to balanced in my preamp,
And then I could just follow your successful values.
I am working on a passive premp with only a Pearl phono with external PS. I could use 4562 with unity gain. but I would brefer not to. |
|
|
| SheldonD |
my last post crossed yours.
I am hanging fire on parts orders because of my confusion,
One thing I can't find is the phoenix terminal posts you are using there are plent at Digikey and mouser but not exactmatch to yours (at least I can't match).
I have spent too long on this alone. |
|
|
| BrianDonegan |
Digikey:
277-1667-ND
277-1578-ND
I would leave the tPM at our standard gain (11) and use the following on TXD:
Rg = 1.1K (matched to 0.1%)
Rf = 22.1K (matched to 0.1%)
That will give you 20X gain in the TXD and you will be very happy. These are standard values in our TXO kit (for TXO 20x). My balanced TXO is running at 10X (Rg = 2.21K). |
|
|
|
