My upcoming LM4780 design

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
It's been half a year since I heard about Gainclones, when I got my FX-120 speakers.

I collected a huge amount of informations on this forum, and surfing the net.

I basically wanted to build the basic LM3875 non-inverted GC, but finally I went for a more ambitious project :

- two Bridged or paralleled LM4780 to build a stereo amplifier
- all P2P connections for ultra-short signal path
- many trials concerning the feedback network and amplifier configurations
- separated high cap. snubberized PSU with 300VA trafo and double discrete rectifier bridge
- fully shielded case with 0.8mm iron plates
- good-looking box

I have right now everything, except the most important part, the LM4780 themselves :D (waiting for the NS samples) and some snubbers, and some RCA sockets.

I begun the assembly with the PSU, as it is the only part where I have all the components.
 

Attachments

  • dscf3818.jpg
    dscf3818.jpg
    54.7 KB · Views: 867
The capacitors are some Panasonic FC 1500mfd/50V. I have ten of them, and will see if I put some on the chip's pins directly or put them on the PSU board. It would give me the equivalent of 7500mfd per rail, I hope it will be enough.

The rectifiers are some MSR1560 from OnSemi : they are ultrasoft, ultrafast (45ns) and quite oversized : they're rated at 15A 600V. I hope they will be good, since I never saw people using them. On paper, they look far better than the popular MUR860.

The assembly is done P2P on a bakelite board with 1mm² OFC cable and electronic solder. I took care to shape the wires in order to achieve the widest contact surface with the pins of the components. I doubt any standard PCB can have this equivalent wire gauge ! I also took care to melt the wire's insulation in the copper strands to prevent them from oxydation (before putting it on the pins of course :D )


Now, a little question : is it better to have all the 1500mfd caps on the PSU board and put some 100mfd local bypass on the chip, or to have some 1500mfd caps left on the chip's pins and some other on the PSU board ?
 

Attachments

  • dscf3819.jpg
    dscf3819.jpg
    65.5 KB · Views: 742
Well, I received my samples from National a few days ago, and I made the first chip run this evening, in a very "prototype" manner (cf. photo) :D

I used thin soldering wire for the connexions, if there was a problem, it would blow very fast and save what could be saved.

I first took care to put myself away from the "prototype zone" in the case it would explode or something else (it's the first amp I build !) and switched it on...

What a relief, nothing happened ;)

I touched everything, to be sure nothing was warming up, and measured the DC offset. Good surprise : only 3.3mv ! And I don't use any capacitor !

I plugged my old speakers, and sent the music : immediately, tighter bass than the old pioneer, and much refined highs. Mids appear clearer, though the speakers are not good.

I finally plugged one of my FX-120 TQWTs. First, the difference was not so obvious, but I heard quite fast some differences : more air, more details, better sound separation. It feels very clean.

I now need to make the other chip run, and hear this in stereo !
 

Attachments

  • dscf3868.jpg
    dscf3868.jpg
    74.9 KB · Views: 550
The design is based on the National bridged amplifier but I changed some values :

Ri is 4.3k
Rf1 is 46.7k
Rf2 is 51k
Rb2 is 4.3k (had nothing left for it)
Rin is 47k

I use no cap as DC bloking cap on the input, and no Ci caps.

I bought 1% metal film resistors, and matched them at 0.1% with a calibrated DMM. (hence the 46.7k resistor, not a commercial value). Maybe it is a reason for the low measured DC offset.

The mute is always deactivated, and it doesn't need a switch as there is absolutely no transcient when swithcing on/off.

The PSU is based on 300VA trafo, MSR1560 rectifiers, and I only use 1500mfd on the chip's pin. I'll try later some other PSU config, to see if it is worth.

The pot is a 10k ALPS log pot, one of the most expensive part :hot: in this circuit. Gotta find cheaper pots if I build gainclones every month :smash:
 

Attachments

  • dscf3852.jpg
    dscf3852.jpg
    51.3 KB · Views: 559
The heatsink is rated 1.4K/W and is isolated from the chip with a silicon pad.

Everything is soldered P2P (quite a long job with the small pins of the LM4780).

The supply rails are soldered on a piece of desoldering braid, it works very well.

On my prototype bench, the chip is fixed on the heatsink with a spring grip.

Everything is star grounded properly on a fast-on connector.
- the signal grounds are first soldered on the ground pins of the chip
- the ground pins are connected on the fast-on with the signal cable ground, the signal cable shield, and the 1500mfd caps ground.

- the fast-on is then connected to the PSU ground with the male connector.

I don't have any hum, and the hissssssssss is hardly audible, even at 2 cm of the speakers. Good surprise compared to the old pioneer's hiss, audible at 1m !
 

Attachments

  • dscf3875.jpg
    dscf3875.jpg
    51.6 KB · Views: 488
Thanks !

I mainly use the desoldering braid because it is flat and has a very wide surface to solder on the pins. Moreover, once "filled" with solder, it is rigid and helps avoiding to break the pins with heavy PSU wire.


I have to make the other one run : it doesn't use the same circuit and is based on "low impedance" feedback network :

Ri = 1k
Rf1 = 11k
Rf2 = 12k

Maybe the source will not like, or maybe it won't be an issue and the amp will sound better. Just got to try :)

In fact the big big problem right now is that I don't have a case at all, not even an old shoe box (which would be dangerous lol ). I drew it on paper but have to build it, or more honestly, make it build by my carpenter friend :D, at least for the wood part.

The heatsink is working lying on the desk, but it only gets warm at body temperature. Hope I won't need the fans I bought :smash: !
 
I have made the second module work today, still in "prototype mode". It is much better in stereo ! :D


It exhibits 2.2mV of DC offset, still an excellent value IMO.

It's hard to tell what configuration is the best, as the chips are still improving with the breaking-in.

I got some fears with it : a not well soldered capacitor made intermittent contact with the PSU rail and caused some lightning ! At first I thought I burnt the IC !


I also tried the 6 other capacitors, and what I can say is that with them, bass is quite rock solid, even on my little FX-120 speakers. Never heard that before, definetely !

But without any snubber, it definetely destroys the highs.

I put some 100nf caps on the chip's PSU pins and on the output of the PSU board.

It does improve the sound but right now everybody is sleeping so I can hardly tell what configuration is the best, low cap, high cap, snubberized, etc...

I have many trials to do with all the schematics of Carlosfm and P-A !!!

Got to get some more little snubber caps :smash:
 

Attachments

  • dscf3885.jpg
    dscf3885.jpg
    64.6 KB · Views: 388
Member
Joined 2002
Paid Member
wintermute said:
Have fun experimenting :) just one suggestion though..... :att'n: some insulation tape or probably better heat shrink around your mains terminal block would probably be a good idea whilst you don't have this in a case, don't want to end up :dead:Tony.

Good point Tony,

I always assume the nephews, neighbour's kids or the cat is going to stick something where it doesn't belong. :D In truth, it would more likely be me after a long day or a few wines. :hot:

regards
 
Lol you are right, I didn't insulate the mains !

the switch is quite far from the experimentation zone, and the only risk would be if someone else would plug my amp and put his fingers on the main switch, which I stricly forbid because my amp is very fragile right now :D

Hope my little sister doesn't know how to plug the main connector in the wall and put her fingers in the switch, or hope my mother won't end doing a crisis seeing all this stuff on my desk (which is scrap in her opinion ;) ) and throw it on the floor, or hope my too curious father won't plug it in and touch everything :D. Hope also that my angry brother won't come with a wood stick and strike everything arround !

Lol no, I don't live in a mad family :smash:

I'll surely put some tape on the main swith, as you suggest it, it is way safer, even if I rarely drink liters of Vodka in my bedroom :hot:


Concerning the amp itself, I've put snubbers on the PSU board (100nf on the connexion between the amp and the PSU) and on the chip's PSU pins (100nf on the 1500mfd capacitors legs).

I let it burn in and try to hear differences between the two schematics I designed. It is quite impossible right now to tell them apart.

I also get a very very small hum noise (audible at 5cm) when both amps are connected on the PSU, it must come from the "temporary" connexion because both module alone don't have any hum.
 
Just another Moderator
Joined 2003
Paid Member
hehehe don't underestimate your ability to do something silly ;) just as greg said it is more likely to be him than anyone else, and I can speak from personal experience, it doesn't have to involve alchohol ;)

on the not hearing any difference IMO if you don't hear a difference straight away, any difference you do hear (later) is quite probably your brain adjusting to the sound of the amp. I had a pair of bolle sunglasses years ago that had very yellowish brown lenses. When you first put them on everything looked the wrong colour. Within a few minutes everything looked "normal" again, your brain compensates for anything that is not "normal". A more extreme example was tests done years ago, where they made people wear glasses that turned everything upside down. after a certain amount of time wearing the glasses everything went right way up again, and was upside down if they took the glasses off!!!

Unless you have two identical amps and do A+B comparisons between the "broken in" one and the just built one IMO you can't trust that any difference you are hearing isn't purely your brain getting used to the sound and tailoring it to the way it thinks it should be :)

Tony.
 
The difficulty to compare them comes from the fact that my speakers are placed assymetrically in my room : my listening room is far from being optimal : first, both speakers are placed at 75cm from the floor, and then, one is placed under my bed (my bed is 1m80 from the floor), so they don't present the same frequency response and bass reproduction.

All I could say is that the amp gets more refined : i heard yesterday things I never heard before in my music : it is not accomodation ! :) (or I hope so).
It's true that you accomodate to sound. But when you hear something realistic, it immediately appears better. Before I had my FX-120 speakers, I had my father's old speakers, and they are not good (veiled, no bass...). When I came back from holiday, I had listened to music through headphones, that are better. And yes, I needed a period to accomodate to the sound of the old speakers. On the opposite, when I got my FX-120 for the first time, they appeared me natural from the 1st second I heard them, and I discovered real "hifi".

There is probably very little difference between the amps anyway. Maybe the one with Ri = 1k has a lower noise floor (less hiss), but I'm not totally sure of that.

Is it normal that my amps get at body temperature (35°C on the heatsink) when idle ? Or is it showing that they are oscillating ?
 
Feedback comparisons

Well, I built a little switch circuit to allow direct comparison between both configuration ( Ri = 1k and Ri = 4.3k). The soundcard was used in mono. The tests where done on only one speaker. I used different music materials (techno, classical, pop, metal...)

I can confirm that there is very little difference between both amps :

- the amp with 1k Ri has :
- a little less hiss noise
- It has brighter and more detailed highs (the difference is still very small)
- voices look a little brighter and feel more alive (very little difference )
- impulse transcients have a bit more impact
- the sound is overall "sharper" and faster
but :
- it sometimes feel that bass is a bit "shorter" and "dry"
- test tones reveal a bit more harmonic distorsion between 10Hz and 40Hz (estimated by ear, my speakers only go down to 45Hz)


- the amp with 4.3k Ri has :
- rounder sound overall, softer and hence a bit less crisp. It sounds more "shy"
- a bit warmer bass
but :
- a little more hisss
- maybe a bit more distorsion above 50Hz
- highs sometimes feel less "complex", the rounder sound makes them a bit hidden


I overall prefer the Ri = 1k version, and I will have to do further listening test to figure out if this version really has less bass (when my parents won't be there :D).

Some people may probably prefer the Ri = 4.3k version, if they have less refined speakers, or don't like "fast" sound.

My Fostex drivers are very revealing and I like sound a bit analytic to hear the best from them, while some would find it harsh.

Hehehehehehehe now, let's beef up the PSU !! :smash:
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.