Rerurn to Romy the Cat's Site

Melquiades Amplifier
Topic: A proper implementation of low-pass filters

Page 1 of 1 (5 items)

Posted by Romy the Cat on 02-18-2008

If to look at the Melquiades schematic in the channel “A” you will see 78H first order low pass filter.

http://www.goodsoundclub.com/PDF/6-Chennal_Melquiades_DSET_Amplifier.pdf

The low-pass filter in Milq and the low-pass filters generally is great subject of interest for me as I do not have final judgment about them. Let me to explain. I did not use e coil – based filter in LF chain of Milq because it would be a very large coil with relatively large DCR. The RC filter is small, the quality of the little cap is very high and the filtration is made by shunting the signal, so the signal does not flow via the cap. Everything is fine, right?

Well, yes and no. I learned long time ago that shunted caps are OK with DC but they are dangers in AC. If you have a more or less LF capable playbacks then make an experiment: run your playback from dedicated, bare and non-adulterated mains and then shunt the mains with an appropriate capacitor of 2-3uF. It will act as a low pass filter but also, no matter how good the cap would be then sound will be changing it the LF characteristic in band pass. In other words the bass will be getting screwed.   So, we the Cats fight with any capacitors in primary…even if it is the capacitors in my neighbor house.

OK, but how about the .068uF cap in the Melquiades channel “A” – it is the same shunted cap in AC line? What I was building the Milq I was planning to experiment with it but I never did. Now to do it would be to r-arrange the Milq biasing and frankly I do not want to make any experiment any more. Still, I have some purely extrapolative proposal that the line filtration by RC method for the channels that handle the lower frequency might be not the most advanced way to go. If someone would make some experiment with it and come to some observation regarding the L vs. RC and their relation to lower bass then please let me know and I will put that 15-20H coke in play as a low-pass filter if I need to.

Rgs, Romy the Cat
Posted by Paul S on 02-18-2008
Plenty of theoretical problems at every turn, but is the R from thick wire really that important?  Likewise, I am not sure of special/worse probems from iron core chokes, although I am aware of hysterisis as a theoretical problem.

I am in no position to compare my own situation to yours, but I have actually had better luck with passive speaker level low pass than with active, albeit this likely has much to do with the way the active was implenented.

Have you not used passive, speaker level L low pass already, Romy?  What other than numbers/theory made you switch, and is it all roses now?

Best regards,
Paul S
Posted by IslandPink on 12-31-2010
fiogf49gjkf0d
Romy
This topic is relevant to me currently as I am attempting to get a 750Hz low-pass into a valve (tube) amp that is driving a bass section ( Supravox 285GMF in an Onken cab ).
I have run through a few options in my head but can see various problems. I look at the implementation you have used in Melquiades bass or bass-mid section, using R/C at the front ; this looks good -  however, if I have a volume-pot preceeding the filter , it looks like the crossover point will be embuggered by the varying input resistance. I can't see exactly how this is dealt with in your own system, but I can see it causing probs in mine.
Are there any other methods you considered but maybe rejected in the context of your direct-coupled amp ?

I have found the Melquiades threads very useful anyway . I'm attempting to blend an Onken (Jensen) bass with a 288-H ( Great Plains ) on Azurahorn . It was going to be 2nd-order on the horn, but now listening to the two with 1st-order crossovers , it sounds much better than I expected . Useful to see that you rejected higher-order crossovers early-on .

Mark
Posted by Romy the Cat on 12-31-2010
fiogf49gjkf0d
Mark, if you have volume-pot preceding the filter that all that you need to do is to put the volume after the filter, of to load the filter to your volume control. The volume control of cause in this case shall not be a regular pot but it shall be a constant impedance volume control. There are an army of L-pads and T-pads that maintain constant impedance for load (L-Pad) of for both source and loads (T-pad). The idea is that if you have voltage divider that give to you your necessary db attenuation then you need to have a contra-divider to balance impedance.  Look for ladder-type attenuators. You can easy find 24-42 step attenuators for any impedance you wish.  If you search at eBay “Step Attenuator Dale” then you will see a lot of Hong Kong resellers who sell them. There are a few circulators and you can get the value and make the attenuators yourself with better switches but I do not feel that it worst to but fingers if ready to do attenuators cost sub $40.

The Cat
Posted by IslandPink on 01-01-2011
fiogf49gjkf0d
Ah thanks Romy , I should have spotted the option of filter before vol pot. This may be enough for now. I do have some Dale stepped-attenuators from a previous project if I need to further experiment .
Now, if I'm calculating, do you know if the vol pot ( eg. 100k to ground ) after the R/C will affect the R/C values ? The grid ( C3m in pentode ) is 'out of the picture' of course.
I can use lower impedances for the R/C bit first to try it out - eg. 2200R/0.1uF . These should be simple to find in good quality parts .

MJ

Page 1 of 1 (5 items)