Potentiometer before White Cathode Follower Output stage in Bass DI?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

genghiscohen

Member
Joined
Dec 28, 2012
Messages
6
I'm working on a Bass DI with a 6SL7 preamp followed by a 12BH7 WCF driving a 5:1 output transformer. B+ is 300v.

dbk4ns.png



Now, the examples I've found using a white cathode follower all have the preceding stage directly coupled to the grid of the upper WCF triode. However, I would like to have the volume pot directly precede the grid of the output stage. The problem becomes how to avoid having the dc offset appear at the pot.

At of the moment I have coupling capacitors before and after the pot, but I worry that I will inadvertently be creating a variable high-pass filter (or some other as of yet unrealized strangeness) as the volume is adjusted. Would a resistor, 470k or so, in series with wiper prevent the cut off frequency from rising to a problematic level? Or am I making something out of nothing?

Thanks,
Nick
 
It should be fine as it is. Just make sure the capacitors are large enough not to significantly affect the low frequency response. I don't know if you just missed it off the schematic, but there should be a resistor from the grid of the lower triode of the 12BH7 to ground.

Cheers

Ian
 
Thank you Ian for your reply.

Yes the Lack of a grid resistor on the lower triode is a mistake in the schematic, my apologies.

My question remains as to how to calculate the size of the two capacitors? Am I correct in treating them as two separate RC high pass filters; the first created by the preamp coupling capacitor and the pot and the second by the capacitor in series with pot wiper and the resistance  wcf grid circuit?

Does the cutoff frequency of the first filter not increase as the resistance between the wiper of the pot and ground decreases? Do the values as shown( .022u and .1u) seem reasonable, or will I have to make them significantly larger as to not effect the bass response?
 
So why not put the gain pot between the tone network and the 2nd stage?

> (0.022u and 0.1u) seem reasonable

Yes, the response changes with wiper setting. Design for worst-case. Pot full up, first cap must drive 330K in sereis with 2nd stage output Z. Call it 330K total. One hi-fi 17Hz hi-pass can be 0.01u into 1Meg so 0.03u in 330K. The post-pot cap only has to drive 500K, but that's awful close to 330k so you may as well use the same value again.

But in a MANY-stage chain, ten 17Hz hi-passes becomes -30dB at 17Hz, -10dB at 34Hz, and about -5dB at 68Hz. This may shave male speech and other bass sounds. Unless you are minimalist, you often aim much lower, even 10X lower. Which suggests 0.3uFd, twice. As that's a large cap, I'd round-down to 0.2uFd then upsize any other place where possible without big money.

2uFd driving a 15K transformer is liable to have a big rise in the 5Hz-30Hz range. You would normally use a bigger (costlier!) cap here. OTOH, you could let it rise and then shorten your other coupling caps to get near flat again. (This was commonly done in 1930s PA.)

 
genghiscohen said:
My question remains as to how to calculate the size of the two capacitors? Am I correct in treating them as two separate RC high pass filters; the first created by the preamp coupling capacitor and the pot and the second by the capacitor in series with pot wiper and the resistance  wcf grid circuit?

Essentially yes, but you also need to take into account the source impedance driving the load R in each case. In the first case the source impedance is the 200K plate resistor in parallel with the tube's plate resistance (say 80K) which is around 60K. This is a very small percentage of the 1M pot load R so we can safely ignore it. The -3dB point with the 0.22uF is therefore around 7Hz - perhaps a bit on the high side. If we change the 0.022 to 0.1 the -3dB point becomes about 1.5Hz - low enough not to affect the overall frequency response.

For the pot driving the WCF, the load is 500K ( the two 1Megs in parallel) but the source is the pot which varies - hence your concern about a variable high pass filter. Now the maximum source impedance of a pot, which is itself driven by a source of zero impedance, is just the pot value divided by four (mid point is the worst case) i.e. 250K. In your case it is slightly worse because the pot is driven from a source of about 60K rather than zero which makes it at most 30K worse - say 280K. This is clearly not a small fraction of the load (500K) so we do have a variable high pass filter. The solution is to make it act well below the frequency range of interest. The worst case is 280K source and 500K load with a 0.1uF cap which has a -3dB point at 2Hz. This is perfectly satisfactory and will be better ( that is lower frequency -3dB point) under all other conditions.

So, to summarise, if you make both caps 0.1uF you should be OK.

Cheers

Ian
 
I usually prefer to put a pot before a gain stage driving the wcf output stage.

Mostly because the wcf needs a decent preceding gain stage due to the less than unity gain of the wcf combined with the stepdown of the output transformer. In the case of 15K:600 which I what I ususally use, it ends up being a divide by 5.3 or so.

So, in order to provide a good strong output signal from the secondary of the traffo, say peaking out at 20vpp or so, that means you need some 100vpp on the primary of the the transformer and some 106vpp or so feeding the wcf.

I'm not happy putting 100Vpp + on a pot, so - I put the pot before a gain stage feeding the wcf+traffo.
Lower output levels, say up to 15Vpp or so at the traffo secondary, no problem.

I did do it the other way once and then changed it to the more usual pot+gain+wcf+traffo chain.

If that's not possible, for reason of not enough gain stages available, I usually use an SRPP instead.
 
Thanks everybody, looks like I've got some interesting options to try out!

First will probably be along the lines of Ian's suggestion of using two .1u capacitors at the pot. I will also experiment with how placing the pot between the tone network and second preamp triode changes things.

If all this fails to clear things up I'll look into the SRPP circuit. I did some preliminary calculations and wound up with 2k for the cathode resistors [ Rk = (2R-Load + R-anode)/mu = (2*15K + 5.3k)/16.5], seems like quite a high value, no?

PRR, is this low frequency rise due to LC resonance? If so, is the idea to move the resonant frequency below the frequency spectrum of interest? By how much?

The transformer I'm using is listed as having a primary inductance of 20H so f = 1/(2*pi*sqrt(20H*2.2uf)) = 24 hz. Closest capacitor I have on hand that would yield a lower resonating frequency is 10u but only rated for 200v, which would seem to be cutting it a bit close. The Jensen schematic I had been working from called for 22u! costlier indeed. Maybe the low frequency bump will have a flattering effect...

I'll try to get some pictures and and sound samples up soon if there's much interest.

Nick
 
The WCF can self-bias. This jacks the input resistance to a very high value. Also the cathode resistor you omitted improves symmetry.

> = 24 hz.  ... Jensen ...called for 22u!

And 10:1 of C is only 3.2:1 of F so 8Hz.

20H at 24Hz is 3000 ohm reactance. If the DCR is 100 ohms and the WCF drives like 100 ohms (total 200 ohms series), secondary unloaded, the Q is 3000/200 or 15! 23dB bump!! Assuming the skirt averages 6dB/oct, it extends almost 4 octaves up, well into midrange!

Now consider wiring 12BH7 as simple common-cathode resistor plate-loaded, parallel. The output resistance is near 3K. Working against 20H at 24H the Q is essentially unity, small bump (few dB). You get some gain: from grid(s) to 600r about 4 (instead of 1/4th). There's no HIGH-voltage heater-cathode strain. PSRR is a wee bit better. It does have to eat significantly more current and power than the WCF to get the same output level, expecially if a low-low-THD number is needed. Just saying.
 

Attachments

  • genghiscohen.gif
    genghiscohen.gif
    2.8 KB · Views: 54
Hold on! there's something wrong here. A 20H winding cannot qualify as 15kohms nominal, unless one considers that an LF cutoff frequency of 120Hz is acceptable.
I believe the 15k winding has an inductance of about 100H, which would put the resonance at about 11Hz with a 2.2u cap. Indeed there is no useful musical information there but all the plucking noise will certainly trigger this resonance. Loading the secondary is also good practice to reduce the amplitude of the hump.
But I would say that as soon as there's a transformer in a circuit, there should be a dependable HPF in the path preceding it. Considering it's a bass preamp, you could calculate one of the aforementioned HP's for a 16-20Hz cut-off, which would make it suitable even for 5-string bass.
Remember that almost nobody has ever heard an electric bass amp that goes down to 40Hz flat. Most of the bass cabs are tuned at about 60 Hz so response at 40Hz is like 6 to 12 dB below.
 

Latest posts

Back
Top