Thoughts On LM1875 Gainclones

[Zero One]

I haven't posted to this forum for a while but I hope this information may be of some help to a few experiemters out there. Don't take this a gospel but rather some ideas I have come up with after about 15 months experimenting with gainclones and particularly LM1875 based ones.

First up it should be noted that I am not trying to run my gainclones as full range, I use a pair of subs, but I do have a speaker system, (well a few in fact) that are very revealing so they help a lot in developing things.

So here are my findings/feelings as a result of too much tinkering to admit to.

1) The LM1875 based clones if not expected to go full range have a sweeter and more musical sound than the other chips, which is why I ended up concentrating on them, they can be made to sound very very valve like and I mean that in the best sense.

2) having tried all power supply methods, there is absolutely no doubt to my ears that battery power gives the best results, utterly grainless, highs that a smooth and very tight control over lower mids and upper bass. any other power supply and all the added complexity now seem a complete waste of time to me, though battery charging must be taken into account.

3) Of the battery set-ups, unless you need the watts the +/- 12V supply (actaully a bit over 13V on well maintained gell cells) sounds best, as you go higher to 24V a certain degree of midrange grain and sibilance is introduced, though it is not bad. By the way using any other LM chip on 12 volts will not yield any more power, actually less in some cases. If you need more power one could try a bridged (dual) 1876 chip config, though I haven't, it should give you about 20w on 12v looking at the specs.

4) The ideal voltage seems to be around 16V but this is near impossible to achieve from SLAs without lots of extra regulation bits that can cause extra problems, I have done it via battery packs from power drills but that is not practical.

5) Miniturising the circuits/feedback sounds better, in the highs and really really good shielding and grounding pays dividends in terms of stability which also effects the ultimate sound quality.
A side benefit of this is the amp runs pretty much stone cold, which must be good for long life.

6) If the shielding and signal paths are done well then running at lower gain is no problem at all, this makes for very tube-like highs, note however this is not ideal for a full range gainclone. I have a pair running on a gain of 16 very sweetly at present.

7) These amps really (let me emphasise that) really need a good preamp, running without one sounds quite average in comparison. Don't even think about going without one!

8) They do not clip gracefully at all, so they are best suited in this config to high efficiency speakers, that is their real place in the audio arsenal.

9) Input caps can make a differnence, but no where near as much as some people think in my opinion, I have had best results by combining some larger and smaller MKTs to give the right value rather than one large cap, I am not keen to remove them as I worry about DC getting into the works.

10) These amp work/sound better it the speaker cables are really really short, ie as amps in active speaker boxes

11) A low gain LM1875 amp will drive piezo tweeters brilliantly, smooth as a babys rear end, providing the piezos are properly set up with crossovers etc so this can be really good for a bi/triamped set up.

12) The limits for supply caps before the actual mid/high range is degrades seems to be:

a) 24V-30V transformer supply around 1000uf

b) 12 or 24V smps supplies around 3600uf

c) 12V battery- doesn't seem to matter much, you could probably use very little, but more has no negative effect as far as I can tell.

13) You would be crazy to expect real bass out of these amps, that is missing the point really, think of these as a kind of solid state valve amp and run the woofers with something else, like a digital amp, for this reason actively limiting the bass signal going throught the amp can be a good idea. My current set up has a very steep cutoff at 57hz, this adds great clarity to the lower mids and upper bass.

14) If you really want to run the amps full range and get bass then you are going to be thinking horn loaded 96db efficient drivers such as fostex, audio nirvana etc, then all should be peachy.

15) This is one amp that input cables do seem to make an audible difference, some cables seem to rob the air out of the sound, I have found thinnish cat 5 based cables work really well, basically keeping the capacitance low seems to be a good thing.

Oh and one little oddity if run on 6v you can build a very nice headphone amp, so long as you limit the output, also tweeters can be run on 6v very sweetly.

I hope this helps a few tinkerers out there, and gives a few ideas.

Zero One

[BWRX]

Hi Zero One. Thanks for sharing your experiences I have a few questions/comments about some of your points.

Originally posted by Zero One
1) The LM1875 based clones if not expected to go full range have a sweeter and more musical sound than the other chips

2) battery power gives the best results

3) By the way using any other LM chip on 12 volts will not yield any more power, actually less in some cases.

5) Miniturising the circuits/feedback sounds better... a side benefit of this is the amp runs pretty much stone cold

6) If the shielding and signal paths are done well then running at lower gain is no problem at all

7) These amps really (let me emphasise that) really need a good preamp

10) These amp work/sound better it the speaker cables are really really short

12) The limits for supply caps before the actual mid/high range is degrades seems to be...

13,14) You would be crazy to expect real bass out of these amps... My current set up has a very steep cutoff at 57hz, this adds great clarity to the lower mids and upper bass. If you really want to run the amps full range and get bass then you are going to be thinking horn loaded 96db efficient drivers such as fostex, audio nirvana etc

15) This is one amp that input cables do seem to make an audible difference

1 - What schematic are you using? There are many different ways to use the 1875, some will sound different than others.

2 - What other types of supplies have you compared batteries to? Any schematics (of a regulated supply you've tried for example)?

3 - Using any other LM chip on 12V could yield more output power depending on the load impedance. The other chips have larger packages and have the ability to dissipate more heat, meaning they can handle a lower impedance load (and supply more outout current) at the lower supply voltages. Some of the chips have the same 5A output current limit so they should have the same output power as the 1875.

5 - Running stone cold isn't a side effect of miniturization but of a good layout. Although it is easier to achieve a good layout when using surface mount components and/or a double sided PCB.

6 - That applies to all of the National chips. I run my LM3875s with the minimum recommended gain of 10V/V no problem.

7 - Depending on the configuration you use, they may benefit from a preamp. Most of the LM chips do like to be driven by a low impedance source.

10 - Minimizing the output capacitance these amps have to drive is definitely a good thing.

12 - I don't believe in ranges of capacitances best suited for a particular ranges of frequencies, but the general rule of using larger capacitance for lower frequencies and higher power is a good one to follow.

13,14 - Your statement is a little misleading because you can certainly "expect real bass" out of these amps. If designed properly these amps will have a flat frequency response down below 10Hz. The perceived "lack of bass" problem is exacerbated by the low output power of these amps, the low sensitivity of common woofers, and the fact that our ears are less sensitive to lower and higher frequencies. Or it could really be that the design you're using is rolling off the low end. Many make the mistake of using a small coupling cap with a low input resistance and wonder why there is no bass. As you mentioned, you are limiting the frequency range to your 1875 amps. This is beneficial for the amps and the speakers because they won't have to "work" as hard.

15 - An audible difference due to input cables can usually be attributed to a couple things - a source with a highish output impedance and/or an amp with a lowish input impedance. Input capacitance and capacitive input coupling obviously also affect the frequency response.

Almost all of your points apply to the other National chips too.

[Zero One]

Hi Brian

Thanks for your comments, I am a little short on time right now but some quick responses.

I have tried a few schematics but in the end the one used for the Aussie Silicon Chip Magazine layout seemed to work best with some small mods, I will try to find the schem and post it later, its pretty std though.

I have tried regular torroids with dicrete and bridge rectifiers, 3500uf smoothing caps, snubber caps. I have also used up to 10,000 uf in the power supply as well, and I have also tried the snubbered power supply (carlos FMs I think, not sure without checking) and 12 and 24 v smps supplies with and without smoothing cap banks and now finally settling on 12V batteries.

I imagine that other chips would go harder into to lower loads on 12v, though I have never looked at this as all my current speakers are easy full range drivers of 8 ohms, which I think suits this chip nicely.

I agree that miniture circuits will not make a cooler amp, but if it reduces oscillation tendencies and improves stability then it should run colder, anyway I have tried all sorts of layouts, this one, which is a hybrid of smds and regular bits P2P is the coolest running of all I have made, it also sounds the best. http://homepage.mac.com/braddles/PhotoAlbum3.html (Note: the bottom one on the copper plate)

The preamp thing is interesting as the changes to the pre are really quite obvious at the amp end, so clearly the amp must be able to faithfully reproduce whatever goes into it without colouration. In fact last night I converted my pre to battery power and this took the sound quality up a couple of notches again. I just wonder how many folks are judging their gainclones potential without really addressing the pre side of things, frankly I can't bear to listen to my clones now without te pre, although in the early days of experimenting I thought it was fine. I do wonder how it might sound with a passive transformer based pre?

I have done lots of playing with caps and the big ones make for better bass, it is very obvious in many cases, but for some reason big caps seem to suck the air out of the highs, with the torroid PS in particular, not such an issue with the smps. I can only guess that the less linear nature of the regular power supply leads to constant large power changes that somehow overshoot the mark as far as what is needed for the highs, hard to explain Brian, I hope you know what I mean, maybe you have some thoughts on this.

On bass I agree with you 100% that the amp should produce real bass, after all the response as you say is very flat, but for me the issue is power, the 6 watts or so that I get from the LM1875 cannot drive the bass speakers loud enough to compete with the efficient midrange/highs I get from full range drivers. I have tried it rolled off at 10hz and it does sound good so long as the speakers are efficient, in the end though I need about 40watts for decent bass output so I use seperate amps and I think it does lead to a smoother more open midrange.
By the way my current bass amps are LM3886 based units with 10,000uf caps running high gain and high voltage and they works really nicely for this.

Once agian thanks for your comments Brian, I really find this area fascinating and I will continue to tweak and tinker.

[BWRX]

You're most welcome for the comments. Thanks for your post! I'm sure others have also enjoyed reading your reflections.

It seems you've tried a number of different types of supplies but haven't used one with linear regulation. If you're looking for better than battery sound I would suggest you give them a shot, especially since you run your amps with +/-12V rails. I would recommend using a transformer with dual 12VAC secondaries, two bridge rectifiers, and two identical LM317 or LT1085 regulator circuits to derive the two regulated rail voltages. I do this for my ~7W class d amps and it works wonderfully. I have yet to try it with the LM3875 amps.

Those P2P photos in that link you posted are pretty cool! Point to point is one thing, but doing it with surface mount components adds a new twist

I use a dual op amp as differential gain buffer for both inputs of a chip amp configured as a differential amp. This effectively makes a higher power instrumentation amp. I mention this because it is similar to transformer coupling the input signal (amplifying only the differential signal input and breaking any ground loops) which is similar to using a transformer based preamp (amplifying the differential signal, breaking ground loops, and volume control).

There does seem to be a bit of a disagreement about how much supply capacitance is best. Personally, I like to provide room for 13mm dia, 5mm lead spacing caps as close to the chip as possible. This allows you to use a maximum of about 820uF 50V caps very close to the supply pins of the chip. I also like to use surface mount components so some 0.1uF ceramics go right next to the supply pins as well. Further down the line (preferably in a separate case) I like to have larger bulk caps (~10000uF) right after each rectifier. If regulators are to be used they would obviously go between the large bulk caps and the on board supply caps of each rail.

As far as switch mode power supplies go, they definitely refresh the supply voltage much faster than the standard transformer/rectifier/cap setup. The LM1875 has the highest PSRR in the 100-200Hz range (at least the negative rail does) which is the frequency that the standard transformer/rectifier/cap setup refreshes at (100Hz for 50Hz mains, 120Hz for 60Hz mains). Above 20kHz the PSRR continues to drop below 60dB for the negative rail and about 85dB for the positive rail. Both of those figures are pretty decent but most switching supplies usually have a switching frequency in the range of 20kHz to 100kHz where the PSRR of the chip is not at its best. That doesn't really address your claim of large caps sucking the air out of the highs, but is just something to note. Larger caps do have higher parasitic inductance than smaller caps simply due to their construction (not all caps are created equally) but that shouldn't have any effect at audio frequencies. In any case, you certainly aren't the first to report this phenomenon.

Please do continue to tweak/tinker and share your findings with us

[Leonidas]

Have about the same experience as you have.
LM1875 is my "personal" best sound .
Tried several configs incl. lm3886 etc. Only left chipamp is lm1875 running low votage on : 12-0-12 ; 160VA tranny.
Sounds the most like my copland tube amp.
Also agree on high efficiency drivers, use fostex fe166e horns on 2x 10 watts and it sounds sweet !!.
Use small BSC on the fostex and, for more control , I use 2x 4400uF, instead of 2x 1500. Took some time to find the right balance between sweet mids and higs and, on the other side, some control in the lows.
I'm very curieus about your battery config??
Any pics and schematics ??

[Zero One]

GDay Leon

Well my current setup for batteries is really temporary, I will be building a more permamant arrangement soon with auto charging etc.

Currently it just consists of 2 12V 7.2AH SLAs, which are wired together from one - to one + terminal, this forms the zero volt connection, a cable is run then from the remaining positive terminal which makes the +12V rail and the remaining - terminal for the -12V rail. Switching is conducted via a double pole switch that breaks the +12V and Zero Volt rail. I am currently using a smart 12V charger and I just disconnect the power leads to the switch and then parrallel charge the two cells together. It runs a surprisingly long time before the Voltage even drops to 12V at which point I recharge.

I now have the same set-up for the preamp, except it also has a 6v battery pack as well which runs the source switching in the amp.

As said I will be building a proper setup which using solenoid switching for the 240V charger power source and has a seperate smart type charger for each battery. The batteries for the power amp and pre will always be seperate, though they could all be run of the one pair of cells.

This is the kit that will be used for the 4 chargers

http://secure.oatleyelectronics.com/...a9e69745398015

I have yet to decide on the 6V one

I may place caps across the battery terminals, some feel it helps but I have not tried to see.

One thing I have found about this amp, and especially on battery power is that it does not tolerate poor sources, all the faults are very much reproduced, there is no vieling mush at all, but with a good source-wow. in my system analogue LPs give the best results, my CD player is just not up to the task, it lacks air and life and my Ipod nano sounds really pretty good so long as the files are AIFF......MP3s just grate! Ultimately I would like to run my CD player on battery power, but thats somewhere down the track.

If you have the chance give the batteries a try, and see what you think a pair of 12VSLAs can be picked up for $40.00 or so and most people have a charger around the house.

[Leonidas]

Thanks for the quick reply.
I'm gonna do it next holliday's ( in may).
Already have a charger.
The + - conection makes the 0 volt ?? what if one cell discharges quicker then the other??
Becomes 0 volt then 1 volt?? How do you stabilise ??
Found out there are several threads on the forum for battery setup.
Have alotof reading to do.
All I needed was another "believer" , like you, to bring my hobby back to life!!
Everybody always talks about other chips, but my strong believe , was that the lm1875 , was something special.
Once I read in an interview that some people like the Shigaraki amp (= lm1875)more then the Gaincard (=lm3875) because it sounded more musical, warmer or some thing like that.
This confirmed my own hearing.

???Are there other believers in the lm1875 out there ???

!! my tiny lm1875 amp which drives some big horns in a very sweet and warm way !!!

[danielwritesbac]

Voltage? I liked my LM1875's on a 36vct. I tried running less voltage, and it stank (smaller soundstage, and it wasn't generous to start with). Couldn't wait to shut that off and return it to high voltage (which was better).

Input filter cap? I prefered it overlarge, running wide open with a very loud 2.5uf mini poly or even larger.

Heatsink? Huge with bar clamp and Artic Ceramique (GC Waldham Type 44).

Pot? Alpa Dual A20k or a fixed L-pad--plain wires kept short, not twisted, not sheilded cable.

Preamp? Not required when applied with enough voltage and heatsink (which is one heck of a lot in both cases).

Maximum? A 38vct will push it up to real audiophile performance (and its utter limits too).

Instead? The un-cramped sound of Thompson chip amps, except. . .

What do I use LM1875 for? The sterility of the gainclone lends itself to wherever a multichannel sound, an effects box, reverb, expander, or a computer source may be in use.
Other amps, which may be more "musical" to start with just can't preserve details when faced with an "effects" source; However, the gainclone's flat blank, lets whatever you want to do with the source shine through with much success.

Impression: A real winner in home theater, room compensations, and, because of the detail, also a winner in very low volume listening, including extra-clear voices on TV and movies.
A big win on efficiency, easily running at 1/2 the amperage of other brands in its class while putting out more power (and consuming half as much as it is advertised to do).

Caveat: A real loser on soundstage, so bad that it draws attention to individual loudspeakers for an un-realistic presentation; however, it will faithfully copy a preamp that does have a generous soundstage, and it will do that cleanly.

[BWRX]

Originally posted by danielwritesbac
Caveat: A real loser on soundstage, so bad that it draws attention to individual loudspeakers for an un-realistic presentation; however, it will faithfully copy a preamp that does have a generous soundstage, and it will do that cleanly.

This sounds like you are incorrectly placing the blame on the amps. My LM3875 amps using the same supply have no problems with soundstaging and imaging. In fact, they come very close to a dual mono setup.

[danielwritesbac]

Oh yes, I've tried dual mono. There's practically no difference.
I was comparing the gainclone to several other amps in my collection. Happily, the gainclone wins out on providing extra-clear TV and movies.
I very much appreciate that at home.

It also has an application as the drive unit for a large PA power amp. You see, when the Aphex units and the reverb units get done, plus the noise in the mixerboard, an ordinary preamp can't make any sense of it. Instead. . .
Run through some fan-cooled (near silent coolermaster) LM1875's instead of the PA preamp, and their clean effect carries right through the big power amp--its like upgrading the whole thing. That's a similar, large scale, exploitation of their skill.

They also make good self-powered monitors when combined with 92db+ drivers. It isn't actually necessary to turn them up much, so on-stage experience is much more pleasant with the gainclone. Even when quiet, they're clearly audiable over everything else. That and they're so darned convenient! Little size, quite durable, only $30 all built and really pack a punch--very cleanly.

I didn't mean to put them down, but I can't retract the cramped soundfield comment because I believe it to be true (and obvious)--unless the gainclone is used with a preamp.