Beginner's Gainclone, HiFi LM1875, The Amplifier Board

output caps

Thanks so much for the info on output caps for protection of speakers. Could you explain to me why it's OK to use polarized caps? I know that it is OK, as the Dyna ST120 that I'm using as a donor chassis has 3300u polarized output caps, but I want to understand why they don't need to be non-polarized. (though I'm certainly glad they don't need to be. Much cheaper this way.) And do they need to have smaller non-polarized caps in parallel with them for high frequencies?

Many thanks,
Keith
 
Unless I'm mistaken your Dyna 120 is single supply, so Polarized caps are fine.
Too sleepy to search for that schematic now.
Now, if you build a split supply amp, they are not needed.
And if for any reason you happen to put them in series with the speaker, they'll have to be NP or Bipolar .
 
Thanks so much for the info on output caps for protection of speakers. Could you explain to me why it's OK to use polarized caps? I know that it is OK, as the Dyna ST120 that I'm using as a donor chassis has 3300u polarized output caps, but I want to understand why they don't need to be non-polarized. (though I'm certainly glad they don't need to be. Much cheaper this way.) And do they need to have smaller non-polarized caps in parallel with them for high frequencies? Many thanks, Keith
A reasonably decent quality parallel pair of 3300u or parallel trio of 2200u does not need an additional treble bypass cap because you already have parallel caps. However, it is okay to add an additional treble bypass cap. Nichicon ES 10u, 4.7u is an example of an excellent performer for this use.

It is extremely popular for maximizing subwoofer amplifiers, due to the easily tunable bass that can thunder instead of boom. No amount of brute force with an equalizer is as effective as harmonic tuning. Works fine with full range amplifiers too. So, there is fun aspect to it.

Blocking subharmonic that the speaker wouldn't reproduce is helpful to both headroom and durability. If the polar cap successfully protects the amp then it also protects itself.

Necessarily, if the model of cap used can survive a bridge rectifier, it should be able to survive speaker output too (especially considering little 25 watt amplifier chips). Likewise, if a cap can't survive being used as an output cap then that model cap is totally unsuitable for use at a power board. Check capacitor datasheet for more complete information. When the amplifier is healthy, there's no DC offset anyway.

However, if the amplifier were to eventually wear out and break, it may be able to damage polar caps, but they do still cost less than speakers. I'll say bipolar caps are preferred.

Optionally, at slightly more cost, you can create a bipolar cap array using multiple polar caps in parallel-series. That prospect is what we'd do for enhancing the durability of output caps if used for a larger size split rail amplifier.

To do the high durability approach with polar caps costs about $8 per speaker, which is quite favorable for costing less than speakers. The 35v caps are quite compact and inexpensive.
 

Attachments

  • bipolar.gif
    bipolar.gif
    45 KB · Views: 5,441
Last edited:
High end sound and sturdy too, for about $3 more.

I wonder if the discussion of speaker protectors might make the more durable Parallel LM1875 look a lot more attractive?

The Parallel LM1875 amplifiers are more laid back due to sending crossover noise into ballast, not speakers. This "makes room" or tolerance for so easily using a NFB-shunt cap (prevents amplification of DC, gives more durability and is more dynamic) while also maintaining a level harmonic response. It won't sound like a typical chip amp. These parallel amplifiers imitate the tone of a rather expensive discrete amp, but also maintain the high resolution of the LM1875 chip. Of course, the Parallel handles more current, without breaking down; so, the parallel is lower risk than the solo.

As with the solo LM1875, again Monoblocs are preferred, not only for wider sound but also to be able to Choose the chip's access to current. Parallel LM1875 simplifies the protection issues when you provide chip current handling tolerance higher than transformer current capacity.
Suggestion: LM1875 monoblocs, not more than 1a worth of transformer per chip or LM1875 stereo, not more than 0.5a worth of transformer per chip. SO, Parallel LM1875 monobloc can use a 2a transformer per each monobloc, and that will sound good. Sort of like this 72va*0.6=43W, which is a bit less than the 50W tolerance of a Parallel LM1875. And, we need that "bit less" for protection. You'll still get your 50 watts on bass peaks from the power supply reservoir/tank section. SO, it will have protection, not shyness.

Also helpful, a center tap style bridge rectifier (conveniently elegant with a pair of Fairchild Stealth Dual) and a CRC power supply, which is a system that flexes a bit (protection) and also runs cleaner (decreased power noise = decreased amplifier workload). We should also consider KeanToken's variable RC from "~" to "~" of the bridge rectifier, snubbing the transformer for much cleaner power. You just turn the dial for whatever is most appreciable for higher resolution audio. That works well with our theme of high end audio done easily. When and if non-audio workload is decreased, then we have proportionately increased capacity for audio power. And, you'll like that.

An interesting prospect for a Parallel LM1875 is to have the pair of low loss 220u power caps in-between the chips so that they both share it. Likewise, it is interesting to use only 1 input cap, only 1 nfb-shunt cap, because there's no need to create differences. Then all of the resistors are arranged as standard 2-resistor mixers. So, they share all of the caps, not the resistors.

Similar to Ashok's stereo board, it is probable that the parallel amp could use 100u||100u (creates a high end, low loss 200u) per each rail at the power circuit for an easier job with capacitor selection.

I would love to see what Ashok would do with a "share all of the caps (evenly) and none of the resistors" Parallel LM1875 monobloc. Possibly, input load resistor can be shared because there's only one input anyway. To maintain the 2 resistor mixer structure, we probably need to add ~470R (input stopper) series to the IN+ pin of both chips.
 
Last edited:
My board cannot take parallel chips. First of all it isn't a very well designed board. It was done in a hurry and there is plenty of scope to improve on it.
I tried some experiments with it and shall post on that later.

Pity I cannot find the board layout. It might be on my 'dead' computer. One hard drive died and the other is still on it. Will have to pull it out to see if it is still there on the good drive. No time for all that now. Too busy.

I'll take some pics of both sides of the board and post that.
Cheers.
 
Applied, in comparison to 1969 JLH amplifier. . .

Speaker protection? Well, that's actually just a good excuse for really rockin your house. Audio Engineers like to say "define the audio band" and that works wonderfully if we do it thoroughly. To me, "define the audio band" means, remove all non-audio workload. For small scale amplifiers, I think that driving a speaker efficiently is worthy of consideration, such as. . .

Compare the New 15W JLH amplifier from 1996 versus the really rockin 10W Classic JLH amplifier from 1969. The new edition did better with a scope but the classic edition did better with a speaker. The Classic has an audio-only workload and progressive current drive bass extension. This difference in output is the output cap, and the Classic has had 44 years of successfully driving speakers. Just 3db more of desired signal is similar to doubling the amplifier power, which is approximately how 15 watts of New booming failed to compete with the Classic's 10 watts of deep, clear, high resolution bass.

Likewise, when adding the output caps to LM1875, I noticed when I cranked up the bass boost on my computer. . . instead of a lot of booming, I got a lot of thunder and what seems to be a lot more audio power, but it really was a lot less non-audio workload in combination with a lovely bit of bass extension.

See also Bob Cordell's website on the topic of bass extension. His method looks elegant indeed, but I think that my methods are more efficient for power and easier to build. Or you can use both for nightclub force. Anyway, it is probably necessary to read about Bob Cordell's speaker port tuning. CordellAudio.com - The Athena Active Loudspeaker
 
Wow, excellent use of inductive series elements as noise filters, for both power and small signal.

Especially noteworthy is the inductive series element between the inverting input and the feedback-shunt resistor, which may drop the gain on RF. Personally, I would have installed a good brand of 1/2w carbon film for feedback-shunt resistor, but, it looks like your layout got the job done.

You might still enjoy a variable RC from "~" to "~" of the bridge rectifier.
The could be a 2u polyester cap and a 50R variable resistor.

P.S.
There's a bit of fun to be had with that board. We'd need to flip the output RC's polarity so that it is resistor first (resistor at amp, cap at ground, is normal) and install a more compact cap. Now, at the output RC, the midpoint between cap and resistor has "Dr. Cherry tap" available. We could cable from there to a resistor added at the other channel's inverting input to install a stereo width effect. To avoid bass reduction, it probably needs a series cap (series with the new cable) and that cap located pin2pin with the output RC's cap (right at the Dr. Cherry tap, of course). Both channels done the same. Now, you have a slightly bass boosting stereo width effect. It is desirable to have the effect so tiny that it is transparently not obvious, so the added resistors would be very high value. Personally, I'd use only enough effect to make it sound like monoblocs, in which case, we may have cancelled some stereo build power circuit crosstalk. This prospect may require an amp that has an NFB-shunt cap present (and perhaps a tiny little polyester cap for treble helper). Adding the NFB-shunt cap would give you a dynamics, safety, power output and midrange boost, and you could set the stereo width effect to conquer the midrange boost. (Adding a rail2rail cap helps too, such as one 250v 4.7u from v+ to v- probably at the two big caps.)
Anyway, it is actually possible to redeploy a very small amount of a popular stereo width effect as a useful compensation on stereo boards. I'd estimate that the effect may remain reasonably transparent up to about 12% of the gain factor or less than that much. It will reduce crosstalk gain by as much and therefore sound slightly quieter (you may need a small amount of regular gain boost to recoup). The circuit tells each chip "don't do what your neighbor does" and a little tiny bit of that can have it sounding like Monoblocs. Generally, negative feedback is responsible for power noise reduction, and I believe it is possible to use a tiny bit of width expander negative feedback on stereo board for stereo crosstalk reduction as well. That board looks particularly well able to do this. The bonus is making leeway in midrange ear tolerances so that you can re-install your NFB-shunt caps for more safety and more dynamics power.
 
Last edited:
Hi folks,

I recently built a K-50 (LM1875) mono kit following Danielcallsbac instructions except that I did not have a variable resisitor for R1 and use the 1K. It sound real good and I can understand the lyrics of the songs! That may be due to the Infinity Primus 153 speaker of course.

I am writing this because I used two Jameco regulated wall transformers, 12V, 2.5A, wired to make +12, 0,-!2V supply. This works very well. At maximum volume it is loud and also clips, but when backed off a bit it is just fine for private listening. This is a light weight and inexpensive solution. Would I prefer 15 or 18 volts? Probably, but this is very satisfactory.