Split plate loads

[clyde1]

What is the advantage of using split plate load resistors for lowering gain versus other methods? Or is there an advantage?

[Chuck H]

Not a whole lot. But it can simplify a design layout sometimes by eliminating extra voltage dividers (good). It can lower output impedance while keeping a higher input impedance on the following grid (really small gains but still a good thing). The series resistance is usually comparably low so there's less top end roll off (also small gains). Probably reduces hiss from higher series resistance used in other gain reduction circuits. Possible hum reduction by reducing the number of ground points that would be used in other gain reduction circuits. I like 'em. And use them a lot.

[clyde1]

Thanks for that Chuck. Is there a rule-of-thumb you use for the ratio of top to bottom resistor vs gain reduction or is it a linear relationship?

[Chuck H]

Thanks for that Chuck. Is there a rule-of-thumb you use for the ratio of top to bottom resistor vs gain reduction or is it a linear relationship?

As voltage gain goes it IS a linear relationship. A 50/50 load split would result in half the voltage gain. Remember that the plate load has an ideal impedance of zero at the HV rail side. So to the AC signal it ideally looks like ground. Whether the linear relationship SOUNDS linear is another matter. Whenever I'm in question on the best ratio I teporarily install a pot as the load. When I find the ratio that sounds right I replace the pot with fixed resistors of the closest ratio. DIACLAIMER: Most pots aren't rated for HV. So I can't endorse doing this just because "I" do. But I can say that when I do it I either turn the amp off between adjustments and allow the voltage to bleed off (checking for safe conditions with a voltage meter) or I use an insulated tool to adjust the pot. Just in case any HV ends up on the pot case.

[Gingertube]

Just a bit of a caution
- using a split anode load you are actually going to get more of any residual B+ rail noise in the signal. So make sure the supply to that stage is squeeky clean.
- You are also going to get lower drive impedance into the next stage so you have more capability of driving it into grid conduction - which you do not want.
Cheers,
Ian

[clyde1]

Would a grid-stopper help with the grid conduction?

[Gingertube]

clyde1,
You may not need it - splitting the plate load means you are driving less signal into the next stage - but if you do have grid conduction on that next stage then a big grid stop (on that next stage) is the way to go. Look at some High Gain preamp schematics and you will often see grid stop values of 47K even up to 220K. They are there purely to limit grid conduction in that stage. They are usually lower gain stages and so they will have low Miller Capacitance and therefore the big grid stop will not cause excessive high frequency roll off - not that rolling off the top is a bad thing in high gain preamps. I could wish that Marshall rolled off a bit more of the top on many of their amps.

That low drive impedance and ability to drive current into the next stage is not unique to split anode load. Where you use an interstage attenuator you will get a similar lower than normal drive impedance but not as low as the split anode load (usually).

Cheers,
Ian

[Chuck H]

Just a bit of a caution
- using a split anode load you are actually going to get more of any residual B+ rail noise in the signal. So make sure the supply to that stage is squeeky clean.
- You are also going to get lower drive impedance into the next stage so you have more capability of driving it into grid conduction - which you do not want.
Cheers,
Ian

I agree on the first point. I havent had any trouble since the rail is usually pretty clean by the time it gets to the preamp. But there are many "vintage" type rail topographies that are under filtered and might cause some noise.

On the second point, I always thought the lower impedance would actually REDUCE the discharge time and help prevent grid conduction. No?

[Steve Conner]

No, because it's a non-linear circuit there are two time constants, one for charging the coupling capacitor and another for discharging.

The charge time constant is set by the output impedance of the previous stage and the grid stopper resistor, but the discharge time constant is dominated by the grid leak resistor.

To mitigate blocking distortion, you can reduce the degree of overdrive, or lengthen the charge time constant, or shorten the discharge time constant, or any combination of the above.

If you want a lot of attenuation, the split plate load is a bad idea. Say you set it up to attenuate by 10x. That applies to the signal, but noise coming from the B+ is hardly attenuated at all. Maybe by 10%, not 10x.

But for a modest attenuation like 2x, the difference between the split load and a "proper" attenuator hardly matters.

[Gingertube]

Steve beat me to it.

Plus to limit blocking distortion you limit the grid conduction current into the next stage by a largish grid stop. That limits how much the coupling cap can charge in the first place.

Of-course some folk like that "farty" blocking distortion in a screaming preamp - Kenny Wayne Sheppard (the "Black and Blue" track) comes to mind.

Cheers,
Ian

[Chuck H]

Thank you gents.

One thing I would mention is that I've used split loads A LOT. One of my designs uses a 5X attenuation on one stage and a 3X on another. I don't notice any hum. What I do notice is very slightly less hiss because I eliminated two voltage dividers and their higher series resistance. As to blocking distortion, be it mild or otherwise, I agree that there are a lot of tonal flavors to be added by operating outside of ideal amplification practices. I suppose I come by most of my favored topographies by trial and error via ear testing. Making sure to stay within safe parameters of course. If it sounds good and operates reliably and safe then it IS good. But following this discussion I am interested in why I don't hear any objectionable rail noise now. I do filter more than average. I use roughly double the uf's on the main filter, all following nodes are typical values though. And I never run more than two preamp triodes from a node in the signal chain. Is there something I should be looking for on a scope or something other than hum I should be noticing?

[Steve Conner]

Why don't you hear any objectionable rail noise? Maybe your rail is very quiet. 5x worse than "very quiet" could still be pretty quiet.

If you get any trouble, it'll be 20 years down the line when the electrolytic rail bypass caps have dried out and gone high ESR. The ESR appears in series with the output side of the split load, so your scheme makes for a 5x greater effect on the tone (which admittedly could still be insignificant )

The ultimate "how not to do it" example of this was the Fuzz Face. It used a split load attenuator on the output stage, with quite a high attenuation ratio, low resistance, and no bypass cap on the supply, so that the ESR of the battery was part of the divider. Quite a significant part too, if you used an old, tired zinc-carbon battery.

[Chuck H]

I just looked up this old (but short) thread: http://music-electronics-forum.com/t1317/

Ray and myself seem to have had similar opinions about split plate loads. And we also both missed the draw back of added rail noise. No such thing as a free ride. I don't see grid loading as a "problem" per se for a few reasons. One would be, as mentioned, if there is a pleasing effect then it's actually a good thing. Ear testing. It'a also not hard to eliminate grid loading with other methods. And last, it's a give and take between the advantages of a wider impedance differential between the output and input. So how to design a particular stage really depends on which side of the see saw your butt is.

For my personal amps I don't see the risk of rail noise being an issue either. But that's because I wouldn't mind replacing a capacitor if/when it becomes problematic. But I will probably stop using higher attenuation split loads for potential commercial designs and I will avoid entirely using them in early or high voltage gain stages.

Thanks you for the reality check

EDIT: I learn so much here. Even when it's just small but significant nuances. I can log this stuff in my brain and improve my pleasure in designing. I just love this place.

EDIT2: A light bulb just went on... It was on the above mentioned design that I noticed what I thought to be poor performance from the Sprague Atom caps. I had very mild ghosting from an amp that I thought had enough filtering. I've been bitching about Atoms ever since. Ha! I upped the uf's and switched to a lower impedance Nichicon cap and all is well. But the problem probably wasn't caused by any difference in the Atom caps. It was possibly caused by the high attenuation split plate load!!! Future experiments will be slow to happen since I don't build all that much. But I'll certainly be paying attention.

[clyde1]

I'm using a 47k/56k split, not an overly large amount of attenuation I'm thinking, 33k grid stopper on the next stage following the volume. Rail noise is not a problem as it's the first stage after over 100uf or so of filtering. Nice to learn the caveats though, thanks gents.

[GainFreak]

If you want a lot of attenuation, the split plate load is a bad idea. Say you set it up to attenuate by 10x. That applies to the signal, but noise coming from the B+ is hardly attenuated at all. Maybe by 10%, not 10x.

Maybe I got lucky but once I needed to attenuate 10x because the output was near 100V so I used 10/100k split load without any problems.
It was a preamp only with 3 stages of filtering (radial Nichicon, 105deg).