diyAudio (
-   Solid State (
-   -   A little bit of advice please (

5th element 30th September 2004 10:08 PM

A little bit of advice please

Over the past few months ive learnt lots about correct grounding techniques etc which were not present in my active crossovers PCB. Now I have changed the speakers midrange from closed box to open baffle and although the set up I had before almost catered for the change its not exactly as Id like it to be so im going to remake it.

This is going to be a preamp and an active crossover, so in other words it has a volume control too. But my main question here is, is it worth making an input buffer for the DAC to drive. At the moment the DAC drives a 50k log alps blue velvet pot. I was thinking about making an input stage using an opamp and setting the input impedance to about 1M ohm so the DAC would have a very easy life driving it. Then the output of the buffer would drive the pot and then there would be another buffer after the pot to drive the active crossover.

This would act a bit like the musical fidelity Xdac tube thingy they first released. Although the output stage in my DAC isnt lacking in drive (OPA627's) and the cable is drives is only 30cm long it cant help to make everything as good as it can be.

Also im not interested in keeping the number of opamps down, the active crossover is going to be full of them, so adding one more at the input doesnt concern me.

Stocker 1st October 2004 12:44 AM

The question is:

Why not?

AndrewT 1st October 2004 01:06 PM

a buffer on the input gives a predictable impedance and might help slightly. Much more importantly a buffer before the set of low pass and high pass filters is a must have. The impedance at the imput of all the parallel filter stages is riddled with caps and needs a beefy output stage with adequate current ability and a low output z feeding into it..
Now a question for you. Equal Value Sallen & Key filters have a gain of 1 or higher. total gain at the end of the string is quite high. the preamp has to be turned really low to get sensible volume after the poweramp. To resolve this I was considering using MFP filters for all the higher gain stages because MFP has an option to choose the gain of each stage. Do you consider it worthwhile? or simply accept low attenuator settings and the resulting slight noise.
regards Andrew T.

richie00boy 1st October 2004 01:40 PM

It looks like your on the right path for a good solid design. However, I would personally not bother in this particular case with a buffer before the pot. My reasoning being that the DAC already has a reasonable quality buffer with reasonable drive capability and 50k is a fairly light load and your cable is short. Adding a buffer will only add to noise and distortion and at 1 meg the input impedance will be high enough to make Johnson noise much worse and be more susceptible to interference.

Make sure you put a cap in series with the pot input, as pots don't like DC at all.

As others have mentioned it is essential to buffer the pot, but I think you knew that already. I use FET input op-amps in this application (to minimise bias current) to keep DC from flowing in the wiper.

richie00boy 1st October 2004 01:49 PM

Regarding the cascading gain issue, I simply use unequal component value Sallen-Key filters as they work at unity gain. The formulas are out there and should be quite easy to find. I wrote a very simple program ages ago to design them. I will put it on my webspace if you want.

Two alternative methods are to use FET input op-amp as a buffer after each stage -- fed from the NFB point of the filter, or attenuate the output of the filter by the same resistor values as the feedback network -- again feeding into a buffer but this time it can be bipolar if wanted.

5th element 1st October 2004 02:06 PM

I was under the impression that johnson noise was only an artifact present with pots, is this not the case?

Also I do not understand why there would be a problem with gain with sallen key filters. All the standard filters (high pass, low pass, notch filter) all have a gain of one. The ones which dont are the ones youd expect to have higher gain like baffle step/ OB compensation and shelving filters. In the active xover I have now (which has a buffer before the filter stages but not before the pot) there is no problem with gain, I have to turn the volume control to 12 oclock or more to get max the max SPL capable from the speakers anyway.

Having thought about it I dont see any reason why not to include it as stocker said, and it will be really easy to take out anyway.

johnnyx 1st October 2004 02:58 PM

I use equal component value Sallen Key filters because it allows independant setting of natural frequency and Q.

When considering the extra gain for my latest active crossover, I thought I could loose some gain by reducing power amp gain. They are chip amps, so they will be stable if the gain is more than 10.

However, I used the free program from Texas Instruments, Filterpro, and this is very versatile. You can choose target Q and frequency for a few types, and it does all the maths for you. So now I'm using unity gain Sallen Key filters. It's a great program.

If an opamp is driving a cable, I use a series resistor in the output, say 150 ohms, so the op amp is isolated from the cable capacitance.


richie00boy 1st October 2004 05:43 PM

Johnson noise is nothing to do with pots, it's all about the impedance of the source -- the higher the more noise.

There are two implementations of Sallen-Key filters:

The equal-components type where all R are the same value and all C are the same value and Q is set by the gain of the op-amp. If you make the gain unity with this type then Q=0.5 which is too low for 4th-order filters as the total Q would end up at 0.25.

The unity-gain type where the ratio of either R (in the case of high-pass) or C (in the case of low-pass) determines the Q.

Swedish Chef 1st October 2004 06:14 PM

Let's talk noise here since there seems to be some confusion: :D

Thermal noise (or Johnson noise) is present in all resistors, under bias or not. The source of this noise is the thermal motion of the electrons. The higher the temperature and the higher the resistance - the higher the noise. But a 1M input is not going to produce more noise than say a 47k input when connected to a source with low impedance (<150 ohm).

Shot noise is present in all forward-biased semiconductor junctions. It is due to the fact that current is made up of individual electrons which jump in a random fashion.

Flicker noise or 1/f noise (this noise has a 1/f frequency dependence) is present in certain components like carbon resistors and pots (not in metal film resistors) under bias. If you do not run any DC current through them they will not exhibit 1/f noise. The source of this noise is not well understood.

Avalanche noise is present in reverse-biased zener diodes exhibiting avalanche breakdown. This noise is quite large in amplitude and therefore zener diodes should be avoided for low noise.

Burst noise or "popcorn" noise will make a popping sound when played through a speaker. Early opamps like the uA709 had large amounts of popcorn noise. The source of this noise is not well understood but high atom number ion plantants (such as gold) seem to increase popcorn noise.

Well, for the moment I can not think of more noise types. Enjoy the ones that are!

If you are building an active crossover the best way to minimize noise would be to use a volume control after the crossover i.e. one control for each channel and band. One way to do this is with a solid state attenuator (such as BB PGA4311). Or a 4-8 gang pot or passive attenuator... If you do this the noise contribution from the opamps is going to be more or less completely negligible.


5th element 1st October 2004 06:14 PM


I use LSPcad for all my active xover design.

All times are GMT. The time now is 03:34 PM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 18.75%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright ©1999-2017 diyAudio