Thanks Andrew.
With the exception of strip board/point board mounting I have used each of these approaches in the past with satisfactory results, even for long term use. I got to this thread because I am interested in prototyping an LME49830/LT1166 combination similar to the one suggested by Bob Cordell in his excellent book (see figure 27.10, page 549), just to find the real-world limits of that approach. In looking for alternatives to the popular IRFP devices, I came across Owen’s Wire Amp project, and was very impressed by the measured results. Not only did I discover the impressive ACD100 series lateral devices (I feel like I must be living in a cave), but I also discovered a very well done, finished LME-based design. Having something that is stable to leverage as the starting point for using the LT1166 to bias the OPS could accelerate my learning process substantially. I don’t really expect to get results that are much better with the LT approach, but maybe it can allow for lower OPS bias current, maybe not. Also, it may be that Owen’s Wire AMP is already at the limit of the LME49830’s performance. While experimenting with the LME-LT idea, I will build the 4 Wire AMPs to update my home system.
Why mess with the LT1166? It’s out of respect for Bob Cordell and his LME49830/LT166 conjecture. Except for Syn08’s LM4702/LT1166 2007 project, which had very respectable results, I found that almost everyone else has dismissed the LT on the basis of theory. Please correct me if I’m incorrect.
More input on the subject is most welcome, and I still want 4 kits from Owen’s group buy.
Regards,
Neel
With the exception of strip board/point board mounting I have used each of these approaches in the past with satisfactory results, even for long term use. I got to this thread because I am interested in prototyping an LME49830/LT1166 combination similar to the one suggested by Bob Cordell in his excellent book (see figure 27.10, page 549), just to find the real-world limits of that approach. In looking for alternatives to the popular IRFP devices, I came across Owen’s Wire Amp project, and was very impressed by the measured results. Not only did I discover the impressive ACD100 series lateral devices (I feel like I must be living in a cave), but I also discovered a very well done, finished LME-based design. Having something that is stable to leverage as the starting point for using the LT1166 to bias the OPS could accelerate my learning process substantially. I don’t really expect to get results that are much better with the LT approach, but maybe it can allow for lower OPS bias current, maybe not. Also, it may be that Owen’s Wire AMP is already at the limit of the LME49830’s performance. While experimenting with the LME-LT idea, I will build the 4 Wire AMPs to update my home system.
Why mess with the LT1166? It’s out of respect for Bob Cordell and his LME49830/LT166 conjecture. Except for Syn08’s LM4702/LT1166 2007 project, which had very respectable results, I found that almost everyone else has dismissed the LT on the basis of theory. Please correct me if I’m incorrect.
More input on the subject is most welcome, and I still want 4 kits from Owen’s group buy.
Regards,
Neel
SMPS Suggestions
Sure, but i would be very happy to see some more feedback towards this direction.
Depending on the preferences, i could design and build a single pcb power supply suitable for all power ranges, for example from 100W to 500W or more, the only difference being the value and size of some parts. Of course the board must be designed to accommodate the biggest power version, which might not be aesthetic or economic (i don't like to see a board with empty spaces for big capacitors or missing parts) but on the other hand the development would be cheaper and would take less time than having several different versions.
The auxiliary voltage, higher than the main voltage should be regulated for best performances. well, this can be done in two ways. the most obvious would be to connect the feedback loop on this voltage and leave the power output, lower voltage just quasi-regulated, which means that at high current consumption this voltage might drop up to few volts. another way is to regulate the main, power outputs using the smps control loop and regulate the aux. using linear post-regulators like it was mentioned few times along the thread. here we might have a problem. this regulators must have a fold-back current limit, otherwise in case of short-circuit, the power dissipation would be huge, tens of watts or more. of course this is is an unlikely event, but must be considered. so, either oversize the regulators parts and heatsinks, or use a current fold-back with delay (allow capacitors to charge without latching to almost zero output by the quiescent current of the driver IC). the easiest way is to use a single large heatsink design where all the transistors and diodes should be placed, the large size could sustain the peak thermal surge in case of aux. short circuit.
Shielding of the power supply would reduce and contain most of the radiated EMI and heavily filtered input would reduce conducted EMI. i might consider using a PFC preregulator, although is not mandatory for audio, will pollute the mains much less and will affect less the equipment connected to the same mains circuit. But i think this can be done in future version, otherwise will take too much time to finish till end of the year.
Sure, but i would be very happy to see some more feedback towards this direction.
Depending on the preferences, i could design and build a single pcb power supply suitable for all power ranges, for example from 100W to 500W or more, the only difference being the value and size of some parts. Of course the board must be designed to accommodate the biggest power version, which might not be aesthetic or economic (i don't like to see a board with empty spaces for big capacitors or missing parts) but on the other hand the development would be cheaper and would take less time than having several different versions.
The auxiliary voltage, higher than the main voltage should be regulated for best performances. well, this can be done in two ways. the most obvious would be to connect the feedback loop on this voltage and leave the power output, lower voltage just quasi-regulated, which means that at high current consumption this voltage might drop up to few volts. another way is to regulate the main, power outputs using the smps control loop and regulate the aux. using linear post-regulators like it was mentioned few times along the thread. here we might have a problem. this regulators must have a fold-back current limit, otherwise in case of short-circuit, the power dissipation would be huge, tens of watts or more. of course this is is an unlikely event, but must be considered. so, either oversize the regulators parts and heatsinks, or use a current fold-back with delay (allow capacitors to charge without latching to almost zero output by the quiescent current of the driver IC). the easiest way is to use a single large heatsink design where all the transistors and diodes should be placed, the large size could sustain the peak thermal surge in case of aux. short circuit.
Shielding of the power supply would reduce and contain most of the radiated EMI and heavily filtered input would reduce conducted EMI. i might consider using a PFC preregulator, although is not mandatory for audio, will pollute the mains much less and will affect less the equipment connected to the same mains circuit. But i think this can be done in future version, otherwise will take too much time to finish till end of the year.
Having designed a regular LTP input stage and a bipolar/fet combo for VAS, I then put together that driver stage as presented in the datasheet for the LME49830. For that, I used two pairs of BD139/140 each with their own current sources and a similar bias circuit.
THD figures are craaazy with that topology, 0.000% 20KHz into 4Ohm resistive, only 0.001% at 40KHz. And that's only in the sim. One thing I noted in this simulation is that it is a bit of an unstable configuration, the compensation cap (Cdom) needs to be rather high, which results in not so nice squarewave slopes, they're relatively shallow. It indeed is a very critical component as the datasheet says. The upside though is that due to the regulation speed there's virtually no over/under shoot and ringing - the tops and bottoms are very very flat and don't exhibit the usual curved corners.
THD figures are craaazy with that topology, 0.000% 20KHz into 4Ohm resistive, only 0.001% at 40KHz. And that's only in the sim. One thing I noted in this simulation is that it is a bit of an unstable configuration, the compensation cap (Cdom) needs to be rather high, which results in not so nice squarewave slopes, they're relatively shallow. It indeed is a very critical component as the datasheet says. The upside though is that due to the regulation speed there's virtually no over/under shoot and ringing - the tops and bottoms are very very flat and don't exhibit the usual curved corners.
I like the idea of keeping the main outputs regulated and using linear regs for the aux. PFC would be nice too.
Other than that, what about supplying one of your existing designs but without any aux supply components?
The SMPS transformer could be supplied with the correct number of aux windings for the required voltage and the AC could be fed to one of Owens PSU boards? That would at least save the cost and space of an extra transformer.
What ever solution you come up with, put me down for at least 4, possibly 6
Probably best to design for the max power this amp can handle, and swap components as necessary to lower the cost. Would the SMPS be very expensive? I think not - and that should allow all to get by with just one board.Sure, but i would be very happy to see some more feedback towards this direction.
Depending on the preferences, i could design and build a single pcb power supply suitable for all power ranges, for example from 100W to 500W or more, the only difference being the value and size of some parts. Of course the board must be designed to accommodate the biggest power version, which might not be aesthetic or economic (i don't like to see a board with empty spaces for big capacitors or missing parts) but on the other hand the development would be cheaper and would take less time than having several different versions.
The auxiliary voltage, higher than the main voltage should be regulated for best performances. well, this can be done in two ways. the most obvious would be to connect the feedback loop on this voltage and leave the power output, lower voltage just quasi-regulated, which means that at high current consumption this voltage might drop up to few volts. another way is to regulate the main, power outputs using the smps control loop and regulate the aux. using linear post-regulators like it was mentioned few times along the thread. here we might have a problem. this regulators must have a fold-back current limit, otherwise in case of short-circuit, the power dissipation would be huge, tens of watts or more. of course this is is an unlikely event, but must be considered. so, either oversize the regulators parts and heatsinks, or use a current fold-back with delay (allow capacitors to charge without latching to almost zero output by the quiescent current of the driver IC). the easiest way is to use a single large heatsink design where all the transistors and diodes should be placed, the large size could sustain the peak thermal surge in case of aux. short circuit.
Shielding of the power supply would reduce and contain most of the radiated EMI and heavily filtered input would reduce conducted EMI. i might consider using a PFC preregulator, although is not mandatory for audio, will pollute the mains much less and will affect less the equipment connected to the same mains circuit. But i think this can be done in future version, otherwise will take too much time to finish till end of the year.
In the past months, I had few customers who ordered SMPS500R and SMPS300R with aux. voltage higher than the main output voltage, they were using the power supplies for similar application. one example is a SMPS500R with main output set to +-54V and aux set to +-65V. the output voltage of the aux supply should not vary more that 2-3% from low or no load to full load, due to cross-regulation. even so, this voltage will be further regulated using a linear regulator for each rail. normally, the voltage of this aux. output can be with about 5.5V, 11V, 16V or 22V higher than the main output voltage. this because the voltage depends on the transformer turns ratio. example: for +-54V main output, the transformer has 5+5 turns for the main output, which is about 11V/turn giving 55V- ~1V forward drop on the diode. then, the aux. winding will generate the same, 11V/turn. we can get +-65V from 6+6 turns, +-76V from 7+7 turns, or ~+-71V from 6.5 turns. I know, is not recommended to wind half turns on any high freq transformer due to flux unbalance which can create this turn, the transformer actually sees two n+1/2 turns, so, a total of 2n+1 turns due to full wave rectification and any unbalance which can be due to uneven loading of positive or negative supply will alter the flux on the legs well below the maximum allowable flux density for which the transformer is designed. (about 0.15 - 0.16 T)
I will try to make few tests in weekend with SMPS500R and SMPS300R to see how will behave in all extreme situations, and for start, i could even provide modified version of these two power supplies, main output in range of +-36V to +-68V and aux higher with about 10-20V where the linear regulators can be cascaded.
based on the feedback received i can design the dedicated power supply.
the cost should not be higher than similar power and complexity existing ones, unless it gets more complex, like extra regulators, EMI filters, housing, PFC.
I even consider to provide the dedicated smps unit for this amplifier project for a special price (not the website one) for those who bought or will buy the wire amp boards, if can gather enough orders to cover the design and manufacture cost, 30-50 pcs depending on features and complexity.
I will try to make few tests in weekend with SMPS500R and SMPS300R to see how will behave in all extreme situations, and for start, i could even provide modified version of these two power supplies, main output in range of +-36V to +-68V and aux higher with about 10-20V where the linear regulators can be cascaded.
based on the feedback received i can design the dedicated power supply.
the cost should not be higher than similar power and complexity existing ones, unless it gets more complex, like extra regulators, EMI filters, housing, PFC.
I even consider to provide the dedicated smps unit for this amplifier project for a special price (not the website one) for those who bought or will buy the wire amp boards, if can gather enough orders to cover the design and manufacture cost, 30-50 pcs depending on features and complexity.
Cristi, I'd certainly be interested in a dedicated SMPS model for this amp. 55/65 sounds good to me, but I could certainly make do with say 45/55 as well. PFC preregulation and ability to use linear regulation for the front end may not be dealbreakers, but will greatly improve the attraction of such a kit, at least to me.
What PF are you aiming for with the PFC in place?
What PF are you aiming for with the PFC in place?
I think a "complete" dedicated PS with higher voltage aux and extra filters/housing for lower noise operation would be of great appeal. If you can propose a couple of designs, you can start an interest list.
Make that 4 kits and 4 boards
Owen,
In scanning back through all the posts I've missed, I see that you are keeping track of boards separately. Since I need 4 boards to go with 4 kits, please make 4 of each.
Are more kits available in this group buy? I understand that the board group buy is closed out. Does that go for the kits as well?
Thanks,
Neel
Owen,
In scanning back through all the posts I've missed, I see that you are keeping track of boards separately. Since I need 4 boards to go with 4 kits, please make 4 of each.
Are more kits available in this group buy? I understand that the board group buy is closed out. Does that go for the kits as well?
Thanks,
Neel
I must amend my thinking a bit. Since I want to use a regulator for the front end, the output from the SMPS into the regulator should be a good 15-20 volts higher than the voltage for the output stage to allow for good regulation. That means my numbers should rather be 55/70-75 and 45/60-65.
This may be obvious, but then again, maybe not.
This may be obvious, but then again, maybe not.
its all closed
and yes Cristi, do keep me in mind, i'll be watching closely to see what you come up with, ive been eating to try out a quality smps and as mentioned paired with this amazing measuring and amazingly compact amplifier this will = win!!
Why not use a full linear supply for the driver section? Putting a linear stage behind a SMPS with the thought of getting a nice clean lin output is going to end up in disappointment. The shunt device will happily pass thru any switching noise unattenuated. MOSFETs are prone to this more so than BJTs although they won't keep out the noise either.
Ok Guys,
Please give me some numbers first: Output power, Main output voltage and current, aux. voltage and current, and preffered size. Is there anyone who wants to sqeeze the amp into a 1RU case ? or a normal height is fine ? i'm looking for an aluminium case to use as housing for this power supply, as soon as i get a rough idea about the required size.
Please give me some numbers first: Output power, Main output voltage and current, aux. voltage and current, and preffered size. Is there anyone who wants to sqeeze the amp into a 1RU case ? or a normal height is fine ? i'm looking for an aluminium case to use as housing for this power supply, as soon as i get a rough idea about the required size.