DIY Solid State (except IC based and ClassA) Amp Thread

Marantz Cinema Series AV Receivers
N

nn_in

Active Member
Joined
Mar 30, 2009
Messages
378
Points
43
Location
bangalore
Is there a consolidated DIY Solid Sate amp thread in this forum for builds done in India.?

Is there a India DIY supplier outside of chip amps?

Which are the DIY solid state kits (pre and power) you would consider both from India and outside of India for 4ohm / 1pair spkrs considering the difficulty in sourcing components

1. ESP
2.Aspen ( I have done the 55N+ ealier)
3. Leach
4. Destroyer

Iam aware of the diyaudio forum which has great info but want to get a India DIY viewpoint.

Thanks
 
Stay tuned - I have been offering Leach low-TIM PCBs based closely on Jens Rasmussen's schematic, but with 8 pairs of output BJTs per channel to DIYers at diyAudio for about 18 months now. Those bare boards are still available, but a bit of an overkill for Indian DIYers - it is rated at 300+ Wpc into 8 ohms at +/- 75V rails.

I started work on a cut-down version of the same, using 2 pairs of output BJTs per channel, with output power reduced to 125 Wpc at +/- 50V rails. It also includes speaker DC protection relay circuitry which is an improvement on an old Elektor design, based on ideas from Rod Elliot's (ESP) site. That layout is completed, but I need to check and recheck it a few times before giving it for fabrication - I'll probably be able to offer the boards by the end of the year.

You should be able to build it with stock locally-available passives and transistors and get pretty good results, or go with premium parts in the signal path and get even better results. However, I'm not sure that it's possible to beat premium chipamp-based composite amplifiers on measurable distortion using purely discrete designs nowadays. Perhaps a composite topology with a discrete power stage can still do it, but it's a lot of work to tweak and improve, while chipamps make it easy.
 
Last edited:
But a good discrete design has good 2nd and third order harmonics (as far as I have seen), and that depth in the sound, rather than a chip based design its a bit thin on that.
 
linuxguru said:
Stay tuned - I have been offering Leach low-TIM PCBs based closely on Jens Rasmussen's schematic, but with 8 pairs of output BJTs per channel to DIYers at diyAudio for about 18 months now. Those bare boards are still available, but a bit of an overkill for Indian DIYers - it is rated at 300+ Wpc into 8 ohms at +/- 75V rails.
Click to expand...

Are there any higher voltage substitutes, or add another 4 pairs of output transistors to lower the SOA voltage across the devices...
The idea is to up the rail voltage to +/- 85VDC... can we use the same PCB, so that i get even more power?

Have couple of 60-0-60 trafos lying... :)
 
Hydrovac said:
Are there any higher voltage substitutes, or add another 4 pairs of output transistors to lower the SOA voltage across the devices...
The idea is to up the rail voltage to +/- 85VDC... can we use the same PCB, so that i get even more power?
Click to expand...

In principle, it should work with 8 + 8 MJL21193/21194 at 85V into 8 ohms. Drivers should also be OK, but pre-drivers may have to be beefed up a bit. The small-signal transistors will have to be changed to things like MPSA42/A92 or similar. You will also have to increase some bias resistors on the small-signal side to keep some quiescent currents the same as before. Also, the layout on the small-signal side is pretty tight, so you have to carefully check the dissipations of practically every component (they're only passing mAs of current, but it all adds up at 85 + 85 = 170 VDC between the rails).

There are also some beefy Magnatec TO-264 devices that have higher SOAR than the MJLs, but the part numbers elude me at the moment. If you're willing to kludge it with TO3s hand-wired to the inline footprint on the PCBs, some more devices are available.

In short, it's possible but requires careful checking and upgrades to avoid magic smoke.
 
Hmmm while I have an appreciation for both subtlety and precision in language, pl treat this as a light-hearted post!

sometimes using terminology calls into question the technical competence of the PCB designer, prompting wannabe builders to general comprehension, He made a serious design mistake which hes been trying to paper over with kluges

Performing hostile kludging of someones active project is always a sensitive issue... could wind up essentially ruining that and can do critical damage to the designers prestige and reputation/standing in the community

If I read between the lines, am not clear if you are sure/confident whether it is possible... but the direction you are showing will not work for me. However, if you decide to bug fix it and host my version: Ive got my asbestos on; flamb me... as an amp builder, always ready for the holy smoke in my eyes.

Edit: A kludge (or kluge) is a quick, messy but functional fix or workaround to a problem; something not built according a design or plan, but assembled from whatever is available to 'make do.'
 
Last edited:
The standard Low-TIM Leach was designed for +/- 58 to 60 V DC rails. The version by Jens Rasmussen with 6 pairs of outputs was designed for +/- 70 to 72 V DC rails. My version is very close to Jens' version, with two additional pairs of outputs to safely allow +/- 75V operation. As you can see, things are already stretched a bit, but +/- 80V should be within reach with minor mods. If you want to go to +/- 85 V, you will have to recalculate dissipations carefully, no avoiding it - it's way out of the specs of both Prof. Leach and Jens Rasmussen's versions.

The main limitation in the higher voltage designs isn't actually the SOAR limitation of the output devices, though that has to be rated carefully. The real problem is that no relay-based DC speaker-protection circuit can reliably break DC currents of 10A at over 50V without arcing and welding the relay contacts closed. You need special high-current inert-gas filled relays to be able to implement a reliable DC protection circuit, and those relays aren't easily available or inexpensive.

I solved the problem at lower rails of +/- 50V DC, using ordinary relays, by using a two-pole relay with contacts in series, thus allowing a safe interruption without arcing. That idea won't work that well at 85V rails - both contacts in series can arc at 40+ volts, 10A - the moment one of the contacts welds closed, the full 85V transfers to the other arc and will melt and weld that contact closed, as well, followed by the destruction of the woofer within a second or so (before rail fuses blow).
 
linuxguru said:
I started work on a cut-down version of the same, using 2 pairs of output BJTs per channel, with output power reduced to 125 Wpc at +/- 50V rails. It also includes speaker DC protection relay circuitry which is an improvement on an old Elektor design, based on ideas from Rod Elliot's (ESP) site. That layout is completed, but I need to check and recheck it a few times before giving it for fabrication - I'll probably be able to offer the boards by the end of the year.
Click to expand...
Hope you made some progress on the project. What is the current status?
 
captrajesh said:
Hope you made some progress on the project. What is the current status?
Click to expand...

The (CAD) layout is completed, but there didn't seem to be much interest, so I didn't tape out the Gerbers for fabrication. Then the rains and flooding happened in Chennai, which disrupted things a bit - the nearest PCB fabricator to me is in Jaffarkhanpet, close to the Adyar river, which is still in spate. So I'll use the dead time now to see if I can improve the layout.

Meanwhile, I got a pair of MX50-SE boards, designed by LJM@diyAudio. These are available as kits and bare boards from EBay vendors in China and HK. The basic design supplied in the kit reportedly has good sonics, but after running some Spice sims, I found some key modifications that reduce distortion (especially H3 and the odd harmonics) and improve stability. Those mods can be kludged into the existing boards, but a fresh layout seems to be the way to go for future builds.

The modified MX50-SE shows simulation results that are as good or better than the Symasym that you've auditioned last year at my place, but has a generally simpler topology with only commodity BJTs (and no difficult-to-find JFETs). Even with upgraded parts, it is likely to be less expensive than the Symasym. LJM has done a very good design at an accessible cost, with reasonable output power (50 to 75W per channel with commonly available output BJTs). With my modifications, it promises to be exceptional. I'll complete the modified build and evaluate the sonics in due course over the next month or so.
 
diesel engine said:
which is the better thermal feed back for the leech low TIM? the original diode or modified transistor sensor?
Click to expand...

The version that I'm using is the Vbe multiplier used by Jens Rasmussen, which has a transistor sensor as well as a 4-diode thermal-compensation chain on the collector-base side of the transistor. It is said to have a slight overcompensating effect, i.e. as the transistor as well as the diode chain heats up, the overall effect on the output-stage quiescent current is to reduce it slightly. It is basically tilted in favour of bias stability and safety (i.e. eliminates the possibility of thermal runaway).

It can be modifed to use a 1-, 2- or 3-diode chain with different resistors in series, and this is said to alter the behaviour to almost neutral or under-compensated. Personally, I'd stick with the proven over-compensated 4-diode chain.
 

Attachments

  • 135-10072.jpg
    135-10072.jpg
    11.6 KB · Views: 49
HiFiMART.com Offers & Deals on WhatsApp
Join WhatsApp Channel to get HiFiMART.com Offers & Deals delivered to your smartphone!

Similar threads

sandeepsasi
My new rig: Torvin Mosfier II Amp + Torvin Euphonious Towers
2 3
Replies
47
Views
16K
alz_tcr
alz_tcr
O
Thinking DIY: DX-AMP aka Destroyer X Audio Power Amplifier
Replies
5
Views
10K
kanwar
kanwar
Back
Top