Mod-286 build thread

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Just found some more detailed Infos on the heat dissipating capabilities of the Modushop Dissipante casings.

http://www.modu.it/CARATTERISICHE_TERMICHE_DISS.pdf

For the 3U / 300mm deep model it says 0.41K/W and claiming to dissipate 230W (4 x TO3-P) in total, that is ca 115W per side.

Is it a good guesstimate that for example a 200W chip amp e.g. based on LM3886 would emmit in the worst case roughy half its power in heat, that is ca. 100W?
 
Last edited:
Data! Fantastic! Thank you.

0.41 K/W @ 230 W -> 0.41*230 = 94.3 ºC rise in temperature, so with a 25 ºC ambient (probably a bit optimistic) you're looking at a heat sink temperature of about 120 ºC. That may be safe for the transistors but it's certainly not safe for the user.

I recommend a heat sink temperature of 60 ºC max. This number is a bit arbitrary, I admit. It's about the highest temperature I can keep my hand on for a few seconds before the pain sets in. For reference, appliance makers are required to maintain 55 ºC or lower on controls and 65 ºC or lower on all other surfaces so my 60 ºC is in the right ballpark.
I use 25 ºC for the ambient temperature. That's a pretty warm living room. 60-25 = 35. The heat sink is specified for delta-T of 70 C and is less efficient (25.7 % less to be specific) at delta-T of 35 ºC. Thus, the max power I'd dissipate in each of the Dissipante 3U heat sinks is:

(60-25)/(0.41*1.257) = 68 W

You have two heat sinks on each side of the chassis, so 136 W total per chassis side. For the full math and explanation see the Thermal Design section of my Taming the LM3886 series.

For the calculation of the power dissipated in the LM3886, see the Thermal Design page I linked to above. For a quick estimate, use 40 W per LM3886 (assuming worst case dissipation - sine wave at Pout = Pmax/2).

Some quick back-of-the-envelope math suggests about 175 W dissipated with a Class AB output stage running at ±70 V at 14 dB crest factor in to a 4 Ω load. That would result in a heat sink temperature of 70 ºC. That's pretty damn hot. Lower the supply voltage to ±60 V and you can shoehorn it in at a dissipated power of 130 W.
So yeah... It'll work (-ish). Definitely don't make the heat sink any smaller.

The Modulus-S will be a discrete output stage with an error correction circuit on it. I expect it to deliver Modulus-level performance. I hope to get 200 W (8 Ω) as that would make a nice marketing blurb. One could build it with, say, ±56 V supply and get 150 W or ±45 V and get 100 W. Right now the design exists in the simulator but not in reality. It needs to meet with reality but that meeting keeps being postponed as other projects pop up on the priority list. It's a constant juggle between "do I want money now" or "do I want the possibility of money later". Sometimes I do need money now so I can keep the lights on.

Tom
 
Last edited:
BTW: The Modulus-S will be a pre-assembled or partially assembled module. I'll keep the schematic and BOM to myself. So it'll be a bit of a culture shift - and a wise business move. More plug-n-play than DIY.

Speaking of heat sinks... I was conducting a heat sink experiment for a prospective client last night. My lab was a cool 18 ºC ambient. I ran two of the channels of my 4xMOD86 amp at 1/3 the rated output power into a 4 Ω load. This is a common test. The two channels were on the same heat sink (12x3.5x1.5 inches - eBay 'BZ4309' chassis; similar to the 2U Modushop Dissipante) and the heat sink reached 45 ºC after an hour. I could actually feel the thermal gradient walking into the room after that hour. :)
I also ran the 32-tone test signal (14 dB crest factor) at just below clipping into a 4 Ω load on two channels. Under this condition the heat sink reached 40 ºC.
A lot of heat is actually conducted away by the 10 mm thick front panel and the 3 mm top, bottom, and rear panels. The Modushop chassis has powder coated steel panels (except for the front which is either 4 mm or 10 mm aluminum) so I'd expect it to run a bit hotter than the 'BZ4309'.

Tom
 
Last edited:
... nothing really exciting happening today as I am laying out the details of the housings ..therefore let's recap some of the experiences over that past week for those how are also new to the diy audio field and want to follow a similar path to a Mod286.

Equipment:
You only need a voltmeter.
No need for an LC-Meter to test the inductors. Just follow the instructions in the from Tom provided description. Also as Andrew and Tom pointed out earlier the values aren't super critical.
Also no need for a function generator either, as the manual give you a link to generate and download a 400Hz test tone, which you then can play from a MP3 player connected to the input. (It might have been a different story if I had run in trouble and would have had to do some error finding).

Soldering:
A decent preferably temp controlled solder iron. I used a 1.6mm chisel tip for the smd work (correction from an earlier post where I mistakenly wrote 0.6mm) and a pencil tip for the diodes and opamps.

LM3886 mounting & heat sinks:
As pointed out above the heatsinks in the pics are too small and for testing only.
Before soldering the two LM3886, I bolted the Mod286 pcb to the heat sink, loosely dropped in the LM3886 (without soldering yet) and also bolted them to the heat sink. Use metal screws (e.g. M4 will work for the metrics guys among you, don't use nylon screws - Tom thanks for the hint). Now the everything is tightened, solder the LM3886s.

Power Supply.
Please find attached a pic of my test setup. It uses both trafo and power supplies from a previous audio project of mine. I also got two boards of the matching Power86, which I will complete later. The manual and Tom describes the selection of power supplies and trafo in detail.
Because it's from an earlier project I'm using a 2x22V, 160VA trafo. Ideal would be a 2x25V, >200VA trafo, which I will get as soon as the amp tests out ok. I am using the Amplimo trafos (from NL) which generally ship cheaper within the EU as the Antek trafos.

Softstart and DC protection:
As you see there is a trafo start up board connector between mains in and trafo. It's most likely not needed for the Trafos of this size. This particular model also serves as a DC protection circuit. However I don't really like it because of the possibly sound affecting two relays and the DC sensing wire running back from the speaker ground. Tom if you can ever design a good circuit for DC protection, I would appreciate the additional insurance for my valuable speakers.

Ah before I forget... use the light bulb as a safety device and current limiter when hooking up everything the first time ... Google if you don't know how to rig it.

... now on with the cabinet selection....
 

Attachments

  • IMG_2065.jpg
    IMG_2065.jpg
    966.3 KB · Views: 376
Last edited:
BTW: The Modulus-S will be a pre-assembled or partially assembled module. I'll keep the schematic and BOM to myself. So it'll be a bit of a culture shift - and a wise business move. More plug-n-play than DIY.
....

Tom

Although I personally regrett it, I can perfectly understand such a decision. E.g. better IP protection, wider customer group i.e. larger market, and probably more efficient from a support/ troubleshooting point of view and more volume... and hopefully somewhat more margin (although sometimes I get the impression, that one is not to make any profit in the diy sector and everything if it's not free it's already too expensive.
 
I enjoy "playing" in the DIY market. I've made many friends that way. I will keep that going and will add circuits as I go along. Some will be DIY and some will be DIy. Yet others may be IDIFY (I Do It For You). Time will tell. :)

Output protection board: That's a common request. It'll likely happen. Same for a transformer soft-start. I'll probably combine the two on the same board.

Tom
 
A lot of heat is actually conducted away by the 10 mm thick front panel and the 3 mm top, bottom, and rear panels. The Modushop chassis has powder coated steel panels (except for the front which is either 4 mm or 10 mm aluminum) so I'd expect it to run a bit hotter than the 'BZ4309'.

Tom

The Modushop Dissipante chassis is also available with 3mm aluminium top, bottom and rear panels.
 
Oh. You're so absolutely right. Thanks. I much prefer the aluminum panels. Whatever steel Modushop uses is hard stuff. Aluminum is so much easier to work with - easier on the tools too.

I've worked with both the eBay 'BZ4309' chassis and the Modushop Dissipante 2U and 4U versions. The Modushop are quite a bit better in terms of build quality. Last I checked the Modushop chassis with steel panels was less expensive than the eBay chassis once shipping charges were included. You can get the Modushop chassis through the DIY Audio Store and through Modushop directly. The latter is especially attractive for those within the EU as Modushop is in Italy.
Modushop will also poke holes in the panels according to your specifications for a very reasonable fee. $35 I think it is.

Tom
 
Last edited:
So as I was busy otherwise last weekend, today I finally ordered the two Dissipante 3U/300 chassis after finishing the DXF CAD file for the rear cut-outs and the holes. The Modubizshop even has a 20% summer discount until the 31st. However they will be on vacation so the casings won't come until end of Ausgust. Oh well...

Tomorrow I can attach the DXF file in case someone is interested.
 
Today I ran a first test with one of the big speakers KEF R700 and not with the old dispensable JBL. I even hooked up the Altair via a XLR connection to check the SPL. I had to crank up the Alttair to about 80% (on an scale from 0 to 100) to get to my maximum desirable listening level. Is that ok?

Remember the test setup still uses the 2x22v trafo per channel and not the maximum possible 2x25v. Also with the Mod286 one can decide very conveniently on the gain of the amp. Several options can be realized by just switching or leaving out two resistors (even only one for the mono Mod286). My two Mod286s are currently built with 20dB gain.

The specs of my Auralic Altair list the max output voltage on the XLR connectors to be 4Vrms at 0dBFS.

The specs for the Mod286 say that for gain option 1) with 20dB gain the input sensitivity is 2V and for option 2) with 26dB gain it is 1V.

Now how do I have to read this information? For example are the 2V input sensitivity for option 1) equivalent to the 4Vrms output, because the 2V input are to be interpreted as +-2V and therefore are 4V peak to peak. Or is 4Vrms the output overdriving the input by a factor two?

Key question: Which gain option is best 1) or 2) for my setup?
 
I run lower gains than most to optimize the gain structure of the signal chain. This has been done for ages in the RF domain. It's optimizing for the lowest noise, thus the best SNR. This does mean that you need to turn the volume knob a bit further to get the same output voltage. No harm in that.

The input sensitivity is the RMS voltage required to drive the amp to clipping.

Tom
 
I run lower gains than most to optimize the gain structure of the signal chain. This has been done for ages in the RF domain. It's optimizing for the lowest noise, thus the best SNR. This does mean that you need to turn the volume knob a bit further to get the same output voltage. No harm in that.

The input sensitivity is the RMS voltage required to drive the amp to clipping.

Tom

I read your words, but sorry I don't understand how to apply your info to my problem.


1) Am I driving the Mod286 with 20dB gain into clipping with the 4Vrms output from the Altair?

2) Or am I driving my amp just optimally as it has a sensitivity of +-2V (in the 20dB version)?

3) If not 20dB, what would be the optimal amp gain for my particular setup?


Sorry for the noob questions.

Thanks a lot.
 
Last edited:
Maybe this is crucial info to the above questions.
The Altair (DAC/Streamer) specifies 4Vrms at 0dBFS on the XLR outputs and 2Vrms at 0dBFS at the RCA outputs.
I would prefer using the XLRs but I have an uneasy feeling that this would drive the Mods286 into clipping. Therefore the questions from the post above.
 
Last edited:
putting 4Vac into a +20dB (10times gain) amplifier will result in an attempt to produce 40vac (56.5Vpk) at the output.
Clearly the amplifier cannot do that.

Tomchr specifies the sensitivity for maximum output as 2Vac and this will give 20Vac (28.3Vpk) at the output.

You control whether the amplifier replays cleanly or clips by setting the vol pot to a sensible level. That control is the responsibility of the operator.
 
Oh boy. This is embarrassing. It's that easy. Thanks Andrew for opening my eyes. I guess I will do some measuring to determine the maxi setting for the DAC output knob. I fear though given the test yesterday that the amp will not nearly have enough power for my speakers :-(. It was set yesterday at 80% (although I don't know if it has a linear scale) and I believe I already heard some clipping while the average sound level was at the upper limit of my listening level.
 
Last edited:
80% rotation means nothing.

Set your rotation to 80% and input a known level of a set frequency (at which your voltmeter is reasonably accurate) and measure the output voltage. That lets you determine the attenuation of that vol pot for 80% rotation.
Full rotation is usually around 270degrees (electrical). 20% less than max is around 54degrees.
Mechanical rotation can be anywhere from 280degrees to over 300degrees.

Having a switchable testing vol pot that steps in 1dB or 2dB or 3dB steps is a useful resource.
I have a switcher that steps in 0.05dB from 0db to -61.05dB (only 12 DPDT switches required)
The most difficult part was getting sufficient accuracy in the -20dB switch resistors that they gave an error of less than 0.03dB, otherwise the 0.05db stepping was useless.
 
Last edited:
putting 4Vac into a +20dB (10times gain) amplifier will result in an attempt to produce 40vac (56.5Vpk) at the output.
Clearly the amplifier cannot do that.

Tomchr specifies the sensitivity for maximum output as 2Vac and this will give 20Vac (28.3Vpk) at the output.

2.0 V RMS, actually. "Vac" is meaningless. Otherwise correct. However, some recordings (like the original, non-remastered, Dire Straits (1985), Brothers in Arms) are recorded such that having the ability to overdrive the amp by a little is really not a bad thing. Sure. The amp will clip on that one sample that actually hits full scale on the DAC, but will be well below clipping otherwise.

There are definitely cases where having some system gain is useful. The Dire Straits example I just mentioned is one of them. I also suspect those who enjoy LPs will need some gain as many phono stages only put out about 250 mV RMS on typical albums. I do maintain that the proper spot for that gain is in the preamp and not the power amp, though.
Put the gain as early in the signal chain as possible for the best SNR. RF engineers have been doing this for decades but somehow it's a novel concept in audio. Go figure. :)

Tom
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.