You're right that when such circuits are sensitive to small capacitances (on the order of pF) it becomes necessary to have a proper PCB made. The boards I have would only require minimal modification to work as a UCD. Tens of pF won't have a huge effect on the circuit so I should be able to get them running with the proper modifications.
When I said 12V supplies I meant +/-12V supplies.
I'll check out the LMH7720 too.
Thanks.
Edit: A google search for LMH7720 turns up nothing. Am I missing something?
When I said 12V supplies I meant +/-12V supplies.
I'll check out the LMH7720 too.
Thanks.
Edit: A google search for LMH7720 turns up nothing. Am I missing something?
BWRX said:
Google is lame on this part for some reason.
Go to National.com
Its one of there new comparators.
Portlandmike said:
That's because there is no LMH7720. It's LMH7220 😱
Just a little number confusion because of the similarity to the MP7720.
Hi Brian,
have you got any more feedback so far? Are you planning a next launch of pcbs, group buy etc?
If not now, would you mind sending me your proven layout/schematics? I'll try to make pcb myself, build it and you'll definetly got some feedback then 😀
have you got any more feedback so far? Are you planning a next launch of pcbs, group buy etc?
If not now, would you mind sending me your proven layout/schematics? I'll try to make pcb myself, build it and you'll definetly got some feedback then 😀
Hi McDark.
I have received listening impressions from two people who are no strangers to Tripath amps and they compared favorably to the Tripath offerings of the same power level. In a couple of weeks I hope to get some feedback from someone who is more of a valve guy. Of course I like their sound very much with reasonably sensitive speakers.
I plan to do another goup buy but want to do some more experimenting before having any more boards made. Since the input impedance of these modules is fairly low I would like to incorporate an op amp buffer into the design. That will solve any frequency response issues with higher output impedance sources as well as give the user an option to try different op amps. I also wanted to work on a +/- regulator board because these modules really require a regulated +/-12V supply and most people would like a simple, compact solution for that as well.
That's easy enough using 3 terminal regulators but I'd like to make a small SMPS to go with these.
I have received listening impressions from two people who are no strangers to Tripath amps and they compared favorably to the Tripath offerings of the same power level. In a couple of weeks I hope to get some feedback from someone who is more of a valve guy. Of course I like their sound very much with reasonably sensitive speakers.
I plan to do another goup buy but want to do some more experimenting before having any more boards made. Since the input impedance of these modules is fairly low I would like to incorporate an op amp buffer into the design. That will solve any frequency response issues with higher output impedance sources as well as give the user an option to try different op amps. I also wanted to work on a +/- regulator board because these modules really require a regulated +/-12V supply and most people would like a simple, compact solution for that as well.
That's easy enough using 3 terminal regulators but I'd like to make a small SMPS to go with these.
Brian,
may I suggest that some common and most cheap solution for SMPS would be MC34063 with ext driver. I've had some deal with it and it looks reasonable. Could be also used as inverter to power whole thing from single polarity supply.
I also wondering considering MP7782 in future designs. As far as I can understand, these ICs would have much better SNR (datasheet says they got about the same in BTL configuration as single MP7720 as opposed to 7731), moreover they are significially better in terms of power dissipation. Bridging also helps increasing PSRR. What do you think?
Also looks like MP7782 could also be used as stereo amp - they all (MP7722,7731,7782) have the same pinout and structure I suppose. There might be a board with BTL/stereo configuration... I could prerer dual mono though 🙂
may I suggest that some common and most cheap solution for SMPS would be MC34063 with ext driver. I've had some deal with it and it looks reasonable. Could be also used as inverter to power whole thing from single polarity supply.
I also wondering considering MP7782 in future designs. As far as I can understand, these ICs would have much better SNR (datasheet says they got about the same in BTL configuration as single MP7720 as opposed to 7731), moreover they are significially better in terms of power dissipation. Bridging also helps increasing PSRR. What do you think?
Also looks like MP7782 could also be used as stereo amp - they all (MP7722,7731,7782) have the same pinout and structure I suppose. There might be a board with BTL/stereo configuration... I could prerer dual mono though 🙂
Hi Brian,
If you ad a buffer ,there is possibility of flexible design for i/v direct connect to dac??
manu
If you ad a buffer ,there is possibility of flexible design for i/v direct connect to dac??
manu
McDark said:
I checked out all three of these chips before choosing the MP7720, and the datasheets show that the MP7782 and MP7722 are basically identical specwise (they have identical pinouts) and as far as I can tell the MP7722 is just a dual version of MP7720 in a thermally enhanced package.
Bridging doesn't increase the PSRR.
nicoch46 said:
Sure, I don't see why not.
McDark said:
You should always bridge when possible. Its an advantage across the board, more power, more headroom, no supply pumping.
I'd not use the MC34063. Its a dinasor, and its got to low of a switching frequency. If you do a SMPS, you really should do bridge too, because otherwise the SMPS will go to zero duty cycle and cause all new issues. Power Supply ripple is better handled by a class D if its low freq, 120hz rather than 100kHz. Higher is better, but high enough to make it not matter.
Switch at 1Mhz and things are more trivial to filter out.
You also don't need as much power as you'd think.
The 7720, 7731, and all of them except the 7782 are basically the same die. Toss in a comparator, like a LMV7239 and you've got game. You can get silly with the comparator and real fast, but its not going to help much. The LMV7239 can turn some really good numbers in a UCD, like 0.00 or better, its not the limiting factor is my point.
As for which part, 7782 looks better, but just use the switches and roll your own. It will likely be better. If I told you the comparator that was used you'd be suprised. Its not anything special at all, but again, the comparator is not important and extra speed is only of merit if the comparator is ideal, and non-are.
The 7782 is based on the MP8040. A decent building block for a bridged UCD. You'll need to get a signal inversion for a bridge. Like a comparator with complementary outputs. Make sure it has no built in hysteresis though.
For SMPS, I'd recommend you do split secondaries. i.e. say unregulated 18 to 24V DC... x2. Then use two buck regulators and make two isolated +12V supplies. Use LC filters to some decent size electrolytics.
Then connect them in series. Done, a solid supply for a bridge class D. If you insist on single ended, then make sure you use a synchronous buck reg, like a MP2303.
If its bridged, I might still use such a part, or something that swiches faster, but minimum duty cycle will be important since a class D draws almost nothing much of the time and its better not to skip cycles. Linear TEch has some great high frequency monolithic bucks that would work well. part numbers are eluding me but something like a LT1376 but with more current would be good too.
It goes without saying that layout in both the SMPS and moreso in the class D will determine the ultimate performance based on experience. Poor understanding of swithing layout pretty much trashes the limits of performance.
PortlandMike
humm..on forum there is a opus dac ,wm8740,with two unit connect direct to new D10.1 a very DUAL MONO system! nice.....
Hi Mike.
I would agree that bridging has its advantages but I don't feel you should always bridge when possible. The other main advantage of bridging is that it gives you the option of using one power supply rail which is simpler than a split rail supply. The disadvantage is that you often have to use a coupling cap in the input stage. There are ways around this though, but you really don't see them being used since a cap can be a cheaper solution.
I'm really curious as to why you keep mentioning/plugging the use of another comparator. Sure it can all but null the hysteresis effect of the comparator internal to the chips but it's more costly and complicates the design as well as the layout.
I don't see how the 7782 looks better than the 7722/7720. I am curious as to what comparator is used at the input of these chips though. Any hints? Maybe something considered low end like the LM2903 perhaps?
From the datasheet I have the MP7782 doesn't seem to be based on the MP8040. The 8040 was supposed to use 100mohm FETs and had higher peak current ratings while the 7782 is advertised as using the same 180mohm FETs as the 7722 and 7720. Plus, the 7782 has almost identical specs and ratings as the 7722. What am I missing?
The route I plan on going, whether it be a switch mode or linear regulator, will involve using a transformer with separate secondaries and regulating each rail separately. The advantage being that the same regulator circuit can be used to regulate both rails. I'm still undecided on whether to work on a buck or a boost converter for a switch mode regulator. They both have their advantages but I already have a buck regulator designed that I'd like to try out. Tight integration would allow me to use the first pair of caps in the CLC filters (what I currently use in the D10.1s) as the output caps for the regulators.
I would agree that bridging has its advantages but I don't feel you should always bridge when possible. The other main advantage of bridging is that it gives you the option of using one power supply rail which is simpler than a split rail supply. The disadvantage is that you often have to use a coupling cap in the input stage. There are ways around this though, but you really don't see them being used since a cap can be a cheaper solution.
Portlandmike said:
I'm really curious as to why you keep mentioning/plugging the use of another comparator. Sure it can all but null the hysteresis effect of the comparator internal to the chips but it's more costly and complicates the design as well as the layout.
Portlandmike said:
I don't see how the 7782 looks better than the 7722/7720. I am curious as to what comparator is used at the input of these chips though. Any hints? Maybe something considered low end like the LM2903 perhaps?
Portlandmike said:
From the datasheet I have the MP7782 doesn't seem to be based on the MP8040. The 8040 was supposed to use 100mohm FETs and had higher peak current ratings while the 7782 is advertised as using the same 180mohm FETs as the 7722 and 7720. Plus, the 7782 has almost identical specs and ratings as the 7722. What am I missing?
Portlandmike said:
The route I plan on going, whether it be a switch mode or linear regulator, will involve using a transformer with separate secondaries and regulating each rail separately. The advantage being that the same regulator circuit can be used to regulate both rails. I'm still undecided on whether to work on a buck or a boost converter for a switch mode regulator. They both have their advantages but I already have a buck regulator designed that I'd like to try out. Tight integration would allow me to use the first pair of caps in the CLC filters (what I currently use in the D10.1s) as the output caps for the regulators.
Brian,
specs are given for reference circuits. 7722 is "stereo" and 7731 is "bridged", generally they are the same. Let's compare noise levels for bridged configuration - 370uV for 7731 and 187 uV for 7782! And - remember you told us? - noise rises in bridged 7720 config. No wonder - 7720 have 190uV being single-ended. That's why methinks 7782 is better. Two times better 🙂
Also thermal resistance for thermally enhanced SOIC is way better - 105C/W for 7720 vs 40C/W for 7782 and other two-ch ICs. But using them as single ended would be not so nice because of channel interference so the only choice is 7782 in BTL 😀
Just keep split suuplies, and you'll get rid of the cap - it's all the same, bridged or not - right?
Personally I'd like to see it bridged - not everyone got such sensitive speakers as you 🙂
Balanced input is also might be useful... Reference circuit in 7782 datasheet looks ok for me, just need to use split supply and no input caps.
McDark said:
I think there is some confusion about the chips we're talking about. I never even considered the MP7731 because its specs were that much worse than the other MPS chips - MP7720, MP7722, MP7782. I was comparing the MP7722 and the MP7782, not the MP7731 and the MP7782. Clearly they are very different!
McDark said:
Yes, the thermally enhanced SOIC has an exposed pad underneath the package that will greatly reduce the thermal resistance of the package if used properly. This package also makes it difficult for DIYers to solder. Channel separation could be an issue with two chips on one die if you wanted to use each half single ended, but they appear to be completely separate so the increased channel interaction would most likely be due to the lack of physical separation between channels (close proximity on the same PCB).
McDark said:
I'm not against the bridged configuration. The extra output power is nice. I like the single ended configuration because you only need one inductor and cap for the output filter. That gives you more room to do a good PCB layout 🙂 Obviously a good layout can be done with a bridged configuration too there are just different compromises involved.
Brian,
there's no confusion,
rather thats the point - I'm pretty sure MP7022 and MP7031 are just the same die as Mike noticed and new 7031 is just a marketing stuff for "powerful" IC. Well, if bridged, we've got doubled noise - that's pretty understandable - it's just summing, and you've seen that effect then bridged MP7020 yourself, don't you? So that's why the specs for MP7031 are much worse - they just given for bridged config and therefore summed noise!
I doubt MPS would make MP7031 with doubled noise for some reason - there's absolutely no point. And that's a key improvement in MP7082 - it has twice lower noise as single-ended, thus bridged it doesn't have as much noise as bridged MP7020/22/31, nearly the same as single-ended them! Moreover, MP7082 in stereo SE would probably have much better SNR than 7020!
I do agree that that case is harder to solder, but not really much than 0805 - just solder diagonal pins first 🙂 And couple of vias beneath the thermal pad could do the trick methinks - some diyers I know solder BGA that way 🙂
My hands are itchning to try it myself... Just got a pair of 7020, but now I see I'll need a 7082 to compare soon 🙂 And some time to make boards for both... 😀
there's no confusion,
rather thats the point - I'm pretty sure MP7022 and MP7031 are just the same die as Mike noticed and new 7031 is just a marketing stuff for "powerful" IC. Well, if bridged, we've got doubled noise - that's pretty understandable - it's just summing, and you've seen that effect then bridged MP7020 yourself, don't you? So that's why the specs for MP7031 are much worse - they just given for bridged config and therefore summed noise!
I doubt MPS would make MP7031 with doubled noise for some reason - there's absolutely no point. And that's a key improvement in MP7082 - it has twice lower noise as single-ended, thus bridged it doesn't have as much noise as bridged MP7020/22/31, nearly the same as single-ended them! Moreover, MP7082 in stereo SE would probably have much better SNR than 7020!
I do agree that that case is harder to solder, but not really much than 0805 - just solder diagonal pins first 🙂 And couple of vias beneath the thermal pad could do the trick methinks - some diyers I know solder BGA that way 🙂
My hands are itchning to try it myself... Just got a pair of 7020, but now I see I'll need a 7082 to compare soon 🙂 And some time to make boards for both... 😀
Hi Yuri. Now you're confusing me with the numbering 🙂
There are four chips offered by MPS:
MP7720: 20W, Mono SE, 9.5-24V, 5A, 0.04%THD+N, 93dB DR, SOIC8/PDIP8
MP7722: 2x20W, Stereo SE, 9.5-24V, 5A, 0.06%THD+N, 93dB DR, TSSOP20F
MP7731: 30W, Mono Bridged, 9.5-18V, 5A, 0.1%THD+N, 90dB DR, TSSOP20F
MP7782: 50W, Mono Bridged, 9.5-24V, 5A, 0.06%THD+N, 100dB DR, TSSOP20F
As mentioned before, the 7722 is a dual version of the 7720. That's pretty easy to see.
The 7731 is the crappy chip of the family. Its specs are worse for whatever reason. No reason to even consider using it when you could use the 7722 or 7782.
Now look at the 7722 and the 7782. Where do they differ specwise? Dynamic range of course. But wait, why do they differ in dynamic range? Well simply because the 7722 is being used in a stereo SE configuration while the 7782 is being used in a bridged configuration.
The noise floor for the 7720, 7722, and 7782 is basically the same, 190uVrms or 187uVrms. The maximum voltage the 7720 and 7722 can swing across the load is 12V because of the single-ended configuration. So the dynamic range is calculated as follows:
20 log [(12/sqrt(2))/(190e-6)]=93dB.
The maximum voltage the 7782 can swing across the load is 24V because of the bridged configuration. Thus the dynamic range is:
20 log [(24/sqrt(2))/(187e-6)]=99.2dB.
Furthermore, the PSRR is the same (60dB @ 1kHz) for all of the chips in their recommended configurations.
Like I said before, maybe I'm missing something but I fail to see the advantages the MP7782 has over the MP7722. Maybe the internal layout was optimized and that was enough for MPS to call it a new part?
There are four chips offered by MPS:
MP7720: 20W, Mono SE, 9.5-24V, 5A, 0.04%THD+N, 93dB DR, SOIC8/PDIP8
MP7722: 2x20W, Stereo SE, 9.5-24V, 5A, 0.06%THD+N, 93dB DR, TSSOP20F
MP7731: 30W, Mono Bridged, 9.5-18V, 5A, 0.1%THD+N, 90dB DR, TSSOP20F
MP7782: 50W, Mono Bridged, 9.5-24V, 5A, 0.06%THD+N, 100dB DR, TSSOP20F
As mentioned before, the 7722 is a dual version of the 7720. That's pretty easy to see.
The 7731 is the crappy chip of the family. Its specs are worse for whatever reason. No reason to even consider using it when you could use the 7722 or 7782.
Now look at the 7722 and the 7782. Where do they differ specwise? Dynamic range of course. But wait, why do they differ in dynamic range? Well simply because the 7722 is being used in a stereo SE configuration while the 7782 is being used in a bridged configuration.
The noise floor for the 7720, 7722, and 7782 is basically the same, 190uVrms or 187uVrms. The maximum voltage the 7720 and 7722 can swing across the load is 12V because of the single-ended configuration. So the dynamic range is calculated as follows:
20 log [(12/sqrt(2))/(190e-6)]=93dB.
The maximum voltage the 7782 can swing across the load is 24V because of the bridged configuration. Thus the dynamic range is:
20 log [(24/sqrt(2))/(187e-6)]=99.2dB.
Furthermore, the PSRR is the same (60dB @ 1kHz) for all of the chips in their recommended configurations.
Like I said before, maybe I'm missing something but I fail to see the advantages the MP7782 has over the MP7722. Maybe the internal layout was optimized and that was enough for MPS to call it a new part?
Brian, yes, sorry, I used 70xx instead of 77xx. But...
You just don't read my postings carefully.
It's not the specs of 7731 are worse. Not only because there's really no reason to make such chip.
Here's my point again:
1. Specs are given for different configuration - SE for 7720 and 7722 and BTL for 7731 and 7782. There's no specs for chip in MPS's datasheets, but rather for its configuration!
Typical specs table for MP77xx looks like:
2. In reality 7722 and 7731 are the same die (Mike confirmed that) and have same specs then not bridged.
3. 7782 has twice lower SNR than 7731 in BTL, and, therefore, twice lower SNR than 7720/22 being in SE.
I's not about 7731 which is not really interesting, no worse no better, but it helps to understand why 7782 is way better than all of them. I beleive we could see soon something like MP7790 soon presented as stereo amp with 95 uV noise being just remarked 7782. Just as 7722/ 7731.
It's not the curcuit with one comparator, inverter and two bridged driver! We have 2 comparators here, so BTLing of these ICs is doubling the noise! That's also why Mike is suggesting using MP8040 and single ext comparator for BTL. But MP7782 could be good enough too.
Mike, wanna say something? 🙄
You just don't read my postings carefully.
It's not the specs of 7731 are worse. Not only because there's really no reason to make such chip.Here's my point again:
1. Specs are given for different configuration - SE for 7720 and 7722 and BTL for 7731 and 7782. There's no specs for chip in MPS's datasheets, but rather for its configuration!
Typical specs table for MP77xx looks like:
And the curcuits are different! SE or BTL!
2. In reality 7722 and 7731 are the same die (Mike confirmed that) and have same specs then not bridged.
3. 7782 has twice lower SNR than 7731 in BTL, and, therefore, twice lower SNR than 7720/22 being in SE.
I's not about 7731 which is not really interesting, no worse no better, but it helps to understand why 7782 is way better than all of them. I beleive we could see soon something like MP7790 soon presented as stereo amp with 95 uV noise being just remarked 7782. Just as 7722/ 7731.
It's not the curcuit with one comparator, inverter and two bridged driver! We have 2 comparators here, so BTLing of these ICs is doubling the noise! That's also why Mike is suggesting using MP8040 and single ext comparator for BTL. But MP7782 could be good enough too.
Mike, wanna say something? 🙄
BWRX said:
You commented that the 7731 had worse specs and there is no reason to use it. Like I said, its two 7720's. If you want to do a bridge and use hysteretic, there are two comparators now. Its tough to get them to switch at the same time. If you use a comparator in front of them, invert one sides inputs, then the problem is gone and your specs will be much better, even as a hysteretic. If you UCD it, the specs will be far better, escpecially at high frequencies.
btw, nulling the hysteresis is imparative if you wish to improve the performance.
I don't see how the 7782 looks better than the 7722/7720. I am curious as to what comparator is used at the input of these chips though. Any hints? Maybe something considered low end like the LM2903 perhaps?
I don't think you can draw my conclusion by looking at the data sheet.
I think your missing the point. Look at the 8040 as a class D building block!!!
If you making an IC that switches power, you are always compromised in the low level analog stuff. Ground just aint ground.... Hence my insistance it may seem, on an outboard comparator. Get it away from mister nasty!
Slit seconds are the right way. Keep in mind if you go single ended, the pumping will be an issue, ruling out most SMPS.
Bucks have serious advantages over boosts. They don't have a RHP pole.
Also, if your into minimizatin, the 7720 will make a quite good buck regulator.
Good Luck!
Mike
Hi McDark. I am reading them carefully.
Mike, where did you get this information from - directly from MPS, a distributor/representative, or did you somehow come to that conclusion on your own?
Look carefully at the datasheets for the 7720 and the 7722. They have the same features, same typical application schematic, same absolute and recommended maximum ratings (except for thermal resistance because of the different packages), same electrical characteristics, same operating specs, almost identical typical performance characteristics, and the same table of switching frequency versus Vdd and feedback component values... What more evidence do you need to conclude that the 7722 is two 7720s on the same die?
I prematurely/incorrectly judged the 7731 based on poor specs without realizing the specs were the result of the circuit topology. The 7731 typical application schematic uses one half of the chip for signal inversion and gain and feeds that output back to the input of the other half which also inverts the signal but only has unity gain. No wonder this has much higher noise than the other configurations... Having realized this, I see that you guys are correct in stating that the 7731 is basically two 7720s on the same die BUT possibly with some sort of limitation because it has a reduced recommended supply voltage range.
We are in agreement here.McDark said:
If those chips use the same die then why does the 7731 have a lower recommended voltage rating than the 7722 (18V vs 24V)?McDark said:
Mike, where did you get this information from - directly from MPS, a distributor/representative, or did you somehow come to that conclusion on your own?
Look carefully at the datasheets for the 7720 and the 7722. They have the same features, same typical application schematic, same absolute and recommended maximum ratings (except for thermal resistance because of the different packages), same electrical characteristics, same operating specs, almost identical typical performance characteristics, and the same table of switching frequency versus Vdd and feedback component values... What more evidence do you need to conclude that the 7722 is two 7720s on the same die?
I prematurely/incorrectly judged the 7731 based on poor specs without realizing the specs were the result of the circuit topology. The 7731 typical application schematic uses one half of the chip for signal inversion and gain and feeds that output back to the input of the other half which also inverts the signal but only has unity gain. No wonder this has much higher noise than the other configurations... Having realized this, I see that you guys are correct in stating that the 7731 is basically two 7720s on the same die BUT possibly with some sort of limitation because it has a reduced recommended supply voltage range.
- Status
- Not open for further replies.