View Full Version : Tube preamp into solid state power amp?

17th November 2007, 08:28 AM
I've recently read that this combination, if done correctly, can sometimes give you the best of both worlds. The tube preamp giving quality musical reproduction with the solid state power amp transfering that sound to the speakers. Does this really work? I would love to have two monoblocks for power amps, but few are powerful enough. You see, my speakers, while of good quality, aren't very efficient and take some pushing. Just seeing if anyone here had any opinions on the subject.

17th November 2007, 02:22 PM
Yes its possible.. all you are doing is taking an input signal and amplifying it. Assuming you can get all your coupling between stages worked out. some guitar amp manufacturers use this ( with greater or lesser success) sometimes so they can market their products as "tube amps"

Master Splinter
17th November 2007, 08:01 PM
Realistically, the effect that people 'like' about the sound of tube amps is actually the way they distort, especially when overdriven - you may not reach this with a tube preamp (or at least not reach it at a level you like to listen to) so you may not gain the sound you are after. Best way to find out is find an accommodating shop and spend a few hours listening to see if it makes a difference worth paying for!

For the power side, try hitting ebay or stores that sell second-hand audio; I picked up an Adcom GFA-555 a few years ago; with 200 watts per channel (or about 32 bazillion watts if you like PMPO figures) its likely to be the cheapest way to drive low efficiency speakers (it can be quite painful with high efficiency ones!)

17th November 2007, 11:04 PM
Since this is for my stereo system, I believe that I am looking for zero distortion. Or, at least as little as possible. I was looking at preamp and power transformers from companies like Transcendent. There are also some good Chinese tube amps out there.

I build all-tube guitar amplifiers, but I'm not sure how to go about building a high quality stereo amp so I thought I would just buy one. It seems that most of the power amps are less than 20 watts per channel. Whoever is using these must have some really efficient speakers. Perhaps loaded horns?

Master Splinter
18th November 2007, 02:21 AM
No, they just overdrive them so they can get the sweet, sweet sound of even order harmonics as the valves distort. If you want zero distortion, you don't want valves.

There was a valve preamp kit in Silicon Chip a while ago (2003) if you want to build one.

Spend a few days listening to various amps and see what sort of sound you like.

19th November 2007, 02:27 PM
I use a tube pre (GK-1) into a SS power amp (AKSA 100N+).....so very nice. www.aspenonline.com (http://www.aspenonline.com) from Melbourne. Both available in kit form (but not Jaycar cheap). Plenty of kit pre and power amps around, both here in Oz and especially abroad. Just have to hunt.

A tube monoblock of the right design will drive inefficient speakers......there are a great many Chinese designed and built tube amps on the Oz market that can deal with high impeadence speakers no worries. Have a yarn to the lads at www.coemaudio.com.au (http://www.coemaudio.com.au) as an example

20th November 2007, 11:20 PM
If you are looking for zero distortion......... a tube amp isnt what you are looking for.:no:

Running a combination of a tube preamp and a modern solid state power amp most certainly is an effective way of achieving a pleasing sound..:2tsup:.... but for minimum distortion...... tubes can not nearly compete with a well designed high end solid state pre amp.

the tube pre can provide that " warmth" & "detail" people claim to hear from tube amps and the solid state amp will provide the power, bandwidth, slew rate, damping factor....... and so on....... that a tube power amp simply can not hope to approach.

I will never argue that a tube amp can be pleasing to the ear and a valid choice for that fact.

Dont ever try to tell me that a tube amp of any discription ( pre, power or RF) can produce the low distortion and lineararity that a properly designed high end solid state amp can.

Excuse me while I duck behind this large road case & await the stunging replies:D:D:D

21st November 2007, 12:47 AM
Stone him! Stone him! :o

Seriously though, I got to thinking a bit more about this..... If you are using this as a audio system amp then most input signals are at a set line level, the preamp (tubes in this case) amplifies this signal and passes it on to the midamp and then on to the power amp stage (solid state in this case) so it could be possible to have the gain adjustable after the preamp stage negating any tube distortion effects........ Then again after a quick glance at some Mckintosh schematics it appears they had the gain adjust on the input side of the preamp.......

21st November 2007, 10:07 AM
now lets look at some circuit configuration issues.

these days with cd players, mp3 players and the like...... in many situations particularly if the power amp has sufficient gain...... there is no need for a pre amp.

straight out of the player into the power amp.... end story.

now if we are looking at amplifiers with professional level input expectations... there is a problem because they are expecting +4dB (V, v, or m...near enough is good enough) about 2 volts for full output.... so a preamp of some sort will be required.
If you have an amp with a domestic input level of -20dB(whatever) about 200mV you are in business without a preamp.

now to locations of volume controls.
many tube preamps will have a volume control usualy a 50K pot as the first component......because it suits tube amp configurations.

most solid state amps will have a gain/ buffer stage before the volume control because that is convienient.

there are a lot of early solid state design that sort of follow tube design conventions..... because thats what designers understood then.

none of these things are hard and fast rules...... designers will do what they see fit for the specific components they are usingh and their design goals.
Mnay times a lot of ( particularly hi fi & guitar) designs employ particular configurations purely as a marketing choice.

if we start talking about vinyl.........OMG.........that is an entire different truck load of worms.

these days we use preamps.......... no they realy arent preamps........to gain controll and add features such as tone controls and source switching.......they are not so much preamps because they do not provide significant gain or processing (like playback curves for vinyl or tape).... but more controll systems.

many of the signal imput conventions that wer still use carry over from valve days........ 50K input impedance and 200mV input sensitivity is what was convienient with a standard valve input stage.

so if you have a valve power amp you can probaly expect to go straight into it from your CD player using the 50K input attenuator as your volume controll.

It is realy all about gain structure.........which is all about noise and distortion.


25th November 2007, 10:21 AM
I'm reading between the lines here, but it seems to me the whole point of the exercise is to run the greatest amount of watts possible into the system to help the inefficient speakers. Hence the leaning towards a solid state amp, where higher wattages are readily available.

My perspective is as a professional musician, where the tube Vs Solid state thing has already spilled into the area where some manufacturers are combining the two, ie Tube pre and SS amp. I'm not going to speculate on which sounds best because it's irrelevant in the context of maximum wattage. However, there is something which is very apparent.......

I terms of maximum usable SPL, a 100W tube amp can keep up with a 400W solid state amp when run through the same speakers. The reason for this is simple and has already been mentioned here. SS amps distorted later, but their distortion is disgracefully horrible and needs to be avoided at all costs, hence the need for bigger wattages for more headroom. The tube amp distorts earlier, but it's a completely different distortion. Initially it comes on like a compression effect more so than distortion, and is actually sounds musical.

SO ciscokid, I'm not convinced your Tube pre/SS amp solution will actually push your speakers any harder. Your 100W SS amp may only deliver the same amount of usable watts as a 20w tube amp.

When you say your speakers are inefficient, what sort of efficiency figures are we talking about?

25th November 2007, 06:14 PM
I would have to dusagree about the tube amp keeping up with a solid state amp 4 times the power.

I understand this is the belief in guitar circles...... but we have to face it guitar amps never have represented stste of the art power amp design AND this is not under the same conditions as in a HIFI context.

The reasons for preference of tube amps in guitar amps differe in many respects for the preferences in HIFI or studio preamps.

The musical instrument market has as much hocus pocus and mumbo jumbo as the HIFI market possibly more.

If we are talking undistorted output power about the best you can hope for from a 4 tube output stage (using commonly available valves) is about 88 watts RMS...... there will be a trade of between bass response and high frequency response......... for good bass response you will need a large output transformer with a big iron core..... this introduced induction which reduces top end response.
additionaly a tube amp will be designed to work into a specific impedance or will be tapped for a specific impedance.
so a a tube amp designed for 88 watts into 8 ohms will do just that.
A solid state amp of 88 watts will almost certainly be designed to run into 4 ohms and will only deliver between 35 and 60 watts ( depending on output stage design) into 8 ohms...... so the comparison is flawed to start with.

now we start talking about compression......yeh tube amps soft clip so the nay seem to be louder when run past their linear reigon...... in gituar applications they are often run well past soft clipping into very heavy overdrive where the output is well distorted ... like 20% distortion would not be unreasonable....... this much distortion will cause an ACTUAL increase in output power delivered to the speaker....... this level of distortion would not be accepted in a HIFI context.

so you want compression...... no problem........ many power amps now come with some form of compression as a descrete circuit section or as part of the output stage design or as part of the power amp chip.
even the uggliest of these compressions will produce a considerable increase in percieved output.

the reasons for using tube pre and solid state power amps are many.
the first has nothing to do with results and all to do with marketing mumbo jumbo.....what possible improvement can putting a tube in a midrange keyboard achieve......

the second would be economics
tube preamps are very easy and cheap to build, no seriously high voltages are required and any coupling transformers if used are relativly small and cheap.
Tube power amps on the other hand are very expensive to build compared to solid state amplifiers of the same power output, both the powersupply transformer and the output transformer will need to be wound to requirement.
If you want to get past the magical 88watts (sorry 100watts) you need to be talking high voltages and exotic valves...... or you need to go to great extents to balance up an 8 tube output stage.... even then you dont get any serious power.
solid state amplifiers from 1 watt up to 200 watts are easily built from kits using commonly available off the shelf parts........ commercial amplifiers these days are so cheap and mnay of them are excelent having rehashed the cutting edge ideas of 20 years ago and banged the out in china.

the third reason is and one that is insurmountable is the performance factor.
You simply can not build a tube amp that is capable of the, noise, distortion, damping factor, bandwidth impedance tolerance and especilay power output of a good solid state design of any size and havn't been able to for at least 20 years.

If you want a power amp that distorst for " artistic reasons" as is required in gituar amplification that is all fine & beaut... but if you want a power amp that is capable of performing to an acceptable HIFI specification ther is no way to build a conventional tube amp that will do it.
Its not tha fault of the tubes..... (you would be surprised at the specs you can achieve from a tube stage with quite some fiddle).... its the fault of what you have to have arround the tube to make it work with other componenets.

Oh I forgot to mention reliability........even with modern manufacturing, tubes arent real reliable or particularly consistent from item to item....... valves wear with age and output declines.....and if you are even half seriuos you need to buy tested and matched valves for output stages.

and lets not forget the size..... any valve amplifier will by necessity have to be huge in comparison to a similar powerd solid state item.

As I said before I can see many good reasons for running a solid stste output stsge with a valve pre amp if thats what floats your boat.......in my view the only reason that stands up for running a valve power stage for HIFI would have to be nostalgic curiosity........


25th November 2007, 07:24 PM
I would have to dusagree about the tube amp keeping up with a solid state amp 4 times the power.

I understand this is the belief in guitar circles...... but we have to face it guitar amps never have represented stste of the art power amp design AND this is not under the same conditions as in a HIFI context.

Great post Soundman but how are you imply I'm a guitarist :) I'm a bass player, and that means that clarity is very important. OK not as important as Hi-Fi, but certainly more important that a guitar amp.

I base my comments on the fact I can get through a typical gig with a 100w tube amp no problem, but need at least 400w of SS power to do the same. I put sound quality aside for a moment because the tube amp sounds nothing like the SS. But the point I was trying to convey is that the relative output from the original poster's inefficient speakers isn't likely to change simply by replacing tube watts with a higher number of SS watts.

26th November 2007, 02:52 PM
Great post Soundman but how are you imply I'm a guitarist :) I'm a bass player, and that means that clarity is very important. OK not as important as Hi-Fi, but certainly more important that a guitar amp.

I base my comments on the fact I can get through a typical gig with a 100w tube amp no problem, but need at least 400w of SS power to do the same. I put sound quality aside for a moment because the tube amp sounds nothing like the SS. But the point I was trying to convey is that the relative output from the original poster's inefficient speakers isn't likely to change simply by replacing tube watts with a higher number of SS watts.

Not guitarist hmmmmmm....:cool:..... thats more a statement of temprament and personality rather than a technical difference:D:D

While it is not popular out side of heavy metal circles to drive bass amps as far into distortion as the six string screamers would, in comparison to HIFI & studio expctations enormous amounts of distortion are tolerable from a bass amp.

In truth there is absolutely no difference between valve and solid state watts if similar levels of distortion are tolerated.

Don't get me started on speaker efficiency.....enough to say that the vast majority of HIFI speakers these days are pittifully inefficient.... hence the need for relativly large amounts of power to drive them.

these days you are hard pressed to find a respectable commercial HIFI amplifier under 80 watts ( don't get me started on claimed power output either) a channel, where not so long ago a 20 or 30 watt HIFI amplifier was considered respectable.

If we are talking about getting some "more schnaps" out of the speakers whilst maintainig a given tolerable amount of distortion... even a percieved amount of distortion replacing a valve power amp with a higher output solid state amp will most certainly do the job.
We have to ensure that
A/ the wattage is measured in the same manner into the same impedance.
B/ that a significant increase in power will be achieved.

so If a valve amp claimed at 100 watts, will typicaly have an actual output of about 80 watts RMS into 8 ohms, prior to the onset of significant distortion.
that would compare directly with a solid state amp with a bipolar output stage typicaly claiming 200 watts into 4 ohms and delivering about 80 watts RMS into 8 ohms prior to clipping.
Now if we were talking about one of the forms of "silly watts" like peak programe or PMPO that amp may be claimed anywhere between 400 and 2000 watts.....as I said lets not go there.

to achieve a significant increase in loudness from either the solid stste or valve amp we would have to move to an amp claiming 800 to 1000 watts into 4 ohms and actualy delivering arround 350 to 400 watts RMS into 4 ohms.

Now if we drive a solid state amp with no compression into clipping even on transients it sounds very uggly very quickly and small amounts of measures distortion sound most unpleasant.

if we drive the valve amplifier into its non linear region especilay only a little and on transients it will yeild measuerable distortions but they will not be as obvious to the ear and will be far less unpleasant.
If we drive a valve amplifier only a a few percent into distortion it will yeild significant increases in percieved loudness along with measurable increases in actual power output......... solid state amplifiers driven as far into distortion would yeild similar measured increases in measured output but the distortion would be much ugglier.

so if we are comparing apples with apples.... AND cooked in the same pie there is no difference between valve and solid state watts...... but the flavour is different.

but if we are comparing apples with elephants... of course valve watts are louder.... but remember you cant put an elephant in your pocket.


24th June 2008, 08:51 PM
I won't argue about what sounds the best, but in the early days they used to run the Amplifiers Class A, now, theoretically speaking, this is where you put a pure sine wave in and get a larger pure sine wave out, the problem was that they ran very hot beacuse the valve was "on" all the time, then they introduced Class AB and Class B and C and various combinations, so that the amp would use less power and you would not need a huge power transformer (in the days before Switched Mode Power Supplies). However the various Classes meant distortion in the output stages,(as some have mentioned, that is desirable to some people), So, in reality and theoretically, if two amps were both run in Class A and one was valve and the other Solid State, you would be hard pressed to tell the difference. But the Solid State amp would need heaps of cooling to continue the same as a Valve Amp. Amos:)

24th June 2008, 11:20 PM
When you say your speakers are inefficient, what sort of efficiency figures are we talking about?

I believe they are rated at 94 db. And, I've always heard that the ratio was 3:1 for tubes over solid state. Not sure which is which though. I do all my guitar playing through real tube amps (no solid state) with tube rectification. I get by just fine with 35 to 50 watts. Give me a pair of 6L6s and I can be heard over the most heavy handed drummer. I am still looking to build a pair of tube amp kits for my home stereo. Lots of kits under 9 watts. I don't think that will be enough for me. The kits over 50 watts, however, cost the world. :no:

25th June 2008, 08:59 AM
Blundering in..

No one here has mentioned frequency response as far as I can see. Valve amps have lousy frequency response as well as more, and more gradual onset, distortion. That offers the light compression that some of these people find so appealing.

Most preamp gain stages are triode based, whereas the output stages in hi fi valve amps can be triode, tetrode or pentode, and of course different valves and output transformers will offer different sounds.

94 dB isn't particularly inefficient. How loud do you want this system to be ? Also be aware that a low power amp with efficient speaks will produce the same sound pressure as a big amp and inefficient speakers, the sound can be more interesting also. The cost of a smaller amp and louder speakers can be lower than building a monster amp.

Plenty of people have coupled 12AX7esque preamps to mosfet output stages. The result is ok.

Remember also the current draw on a class A stage is constant so you can't physically get sag on that circuit. I use valve rectifiers on class A stages but only because I'm too lazy to use standby switches.

Just some thoughts.

Edit: Just a couple of quick additions. Soundman I understand your trying to avoid writing a text on this but a couple of things you say are potentially misleading:

1. There is a lot more to output transformer frequency response than size.

2. KT88/90/6550B output valves are relatively common and will produce rather more than 88 watts at negligable distortion (in a quad). What is neg dist is relative of course. 0.5% THD in my books. Balancing a quad, sextet or octet of output valves isn't difficult. Nowdays there seems to be a cult built up around valve matching/tuning/boutique supply. Perhaps todays valves are very much inferior to the old ones, or perhaps marketing people have found a new gullibility to exploit.

3. Valves wear out if they are driven hard. I have amps from the 60's with all their origional valves, not so thelectroytics. Carbon resistors and some caps are much less long lived than valves as are some early semiconductors.

and finally regarding power and distortion figures, in the old days there used to be common standards, then marketing people discovered they could publish misleading numbers from dodgy tests. Happened with cars also. Very frustrating. Perhaps the answer to the origional question needs to begin with better definition of what is trying to be achieved ?

25th June 2008, 01:25 PM
Danian there is much truth in what you say.

as to the frequency response.
The frequency response in valve circuits generaly is not down to the valves, many valves used for audio applications will perform just fine up into the low RF frequencies.
The limiting factor on valve power stages IS the transformer, and size is an unavoidable side effect of the engineering.
to get good low frequency performance out of a valve output stage you need a transformer with a large core of iron so it does not saturate thus you need a large amount of wire and there will be a significantly large inductance that will limit the high frequency response.
I had a rack of amps out of one of the cinemas here at one stage the output was a pair of KT66's and they did about 20+ watts into 8 or 16 ohms.
I measured the response of those amps as running from 16Hz up to about 8.5KHz. and yess the transformers were quite large, the output tranny's were twice the size of the supply tranny's.

now most guitar amps need nowhere near that sort of low frequency response so the transformers are significnatly smaller, frequency response can be expected to be more like 80 to 12.5K. That is no problem to a guitarist as the isntrument wont produce out of that range and besides most guitarists cant hear past 10K.:;

Now tarsformer design is much more complicated than that, but the reality is that there would be very few valve output tranny's that could be considered modern in design let alone cutting edge....and the inductance caused bandwidth problen will remain.
The only reason that the achieved bandwidth is available is only due to the use of negative feedback in the amp.
Interestingly in the cinema amps it was necessary to change the feedback as well as the impedance tapping otherwise the top end frequency response suffered very significantly.

Indeed most of the preamp stages in most plain audio amps are triodes or more frequently dual triode valves.....this is because that is all that is required.
A properly built triode preamp will perform to the capacity of the valve. A valve preamp can be very quiet, low in distortion:;, reasonably linear :;, and have quite respectable frequency response.
Oh and they generlay operate "A" class.

In a power stage it is necessary to insert extra plates into the valve to exert better controll of the current flow and thus improve performance in many ways in particular distortion. It is the controll of relativly high currents that valves struggle with. Transient response is the problem, achieving frequency resoponse isnt soo much of a problem (well less than the transformer) but achieving good distortion at the same time as high frequency rsponse is.

yep 94dB/W isnt too bad for a HIFI speaker, it would be OK for a pro audio speaker, I have seen HIFI speakers down in the low 80's and mid 80's is very common.

Yep there are plenty of 12AX7 & similar preamps out there combined with all sorts of solid state power stages, and as stand alone preamps for recording and the like......If its a valve preamp sound you want the result is pretty hard to beat, unless you start looking for esoteric valves.
Cheap & easy to build too.

As to the output of various valves.
if we are talking similar levels of distortion.
lets start with a reference of .01% which in a solid state amp is considered prior to onset of clipping, assuming no rail sag or other misbehaviour.

5% is a significant amount of distortion and will produce more measured power.

even from a valve amp 5% sounds distorted, if you are playing guitar or other instruments it will sound fine:roll::2tsup::no::;.
but if you were playing quality speach or pure type sounds it will sound distorted.
At one time 1% would have considered " respectable " for consumer electronics and would problay be considered listenable for a valve radio.

but remember I am trying to give an aples & apples comparison.
At 1% most solid state power stages will be starting to sound nasty, well at least anoying. Some solid state amps will not tolerate being driven near or into clipping for any length of time. The old 480 module ( and its relatives ) is a good example.

So if we look at typical 4 valve kT88/ 6550, output stages they will be sold as 100watt and will output a genuine 88 watts before clipping.
Now of course you can push the friendship by selecting valves and bumping up the rail voltage but thet is a....... um........ "short term" solution.

Balancing valves is an important & significant matter, I would go further batch testing of valves is and always has been a significant matter.
Because of the crude and mechanical construction of valves the performance can vary substantialy from item to item. In the hey day of valves most sellers had a valve tester and many valves came in packaging that allowed the valve to be pluged in for testing while the box was still sealed. Many "respectable" valve retailers would not let a valve out of the shop untested.

Some of the top shelf old valves were very much better than others, I would not say generaly thet modern valves as a whole are better or worse, but I would expect that the range of quality has narrowed. The realy top shelf stuff will be long gone, but likewise the inconsistent and unrelible cheapies will also have passed.

Yep... Valves wear out......end story.
This is another problem particularly with output stages, the power output can only be guaranteed with valves with close to spec emission.
as the valve wears out the power output and other performance will degrade.
It is not unreasonable to expect an amplifier will still be functioninf but delivering less than 50% output.

As for the expressions of power and distortion:~:no::doh:, you could do a long thread on that one alone.
The important thing is to make sure the poweroutput is expressed in the same trems and under the same conditions.

100 Watts RMS continuous, pink noise, both channels driven, at .001% distortion, into 4 ohms at 25 deg C.
Now that is a speck you have difficulty arguing with.

As for improving the performance of a tube output stage, the simplest thing is to replace the tube rectifier with a solid state one, there will be much less rail sag and better transient power output.

more later. cheers

25th June 2008, 01:43 PM
I wrote 0.5% but it may not have been clear.

Mass is required to avoid saturation, but interleaves, winding density and geometry affect frequency response, albeit more suttly (sp).

Extra elements affect efficiency.

600V on a KT88 doesn't bother it.

I've got a valve tester, and a big box of NOS valves. I've been at this a long time. I can assure you an amplifier with conservative voltages will run with a set of quality valves for many many years, but whatever. Doesn't matter.

We're probably saying similar things in somewhat different ways.

25th June 2008, 06:31 PM
Indeed what you say is true.

the limitations are and have always been not with the valves themselves but with the surrounding components, predominantly with the transformers.

It is a rare and expensive valve amp that will have anything like a "state of the art " output transformer.
As the quality and the voltage involved in both the output and supply transformers increases the cost skyrockets.

Cosidering that the transformers is where an overwhelming majority of the cost of a valve power stage is, I can understand this is where the corners have always and will always be cut.

From what I gather most of the common valve amps run somewhere arround 350 to 450 volts in the output stage which is a long step down from 600V.

one of the other limitations is getting capacitors ( particularly now) that are suitable at those voltages.

All these and other reasons are why a valve power stage is much less viable than a solid state one.

so running a valve pre-amp and a solid state power amp is both a reasonable technical and cost option for someone wanting a " smoother more melow":roll: sound.


Isaac Hunt
25th July 2008, 09:22 PM
Over the years I have used many different preamps.Transistor,chip,battery powered chip,passive ,TVC and valve.
There are good and bad preamps of every type.
By far the best preamp I have found is my current one.This is a Supratek Cabernet which uses 300B valves.I use it with both valve and SS power amps in a triamped speaker.
Modern well designed valve preamps are low distortion and wide bandwidth with perfectly flat frequency responses.Valve technology has come a long way in the last 10 years-whereas SS amps have gone nowhere[maybe with the exception of chip and Class D type amps].You cannot compare new valve amps with old guitar amps.
The distortion factor is a bit of a myth.What is the point of .001 % distortion when your speakers are probably at least .5% distortion?Some of the worst sounding amps you can hear have ultra low distortion.

17th July 2009, 04:18 AM
If you want to experiment with a high quality tube output stage, look into an ultralinear or Williamson output configuration. With a good transformer this output stage will provide a flat frequency response from about 15 hz up well past 100 Khz, albeit at the expense of some output power. This is accomplished by using screen taps on the output transformer providing negative feedback to the screens forcing the outputs to behave in a very linear manner. Damping factor of course isn't as good as solid state, but still pretty respectable for 55 year old technology.

The majority of the "Tube" or "Valve" sound is created in the output stages of the power amp. A good tube preamp stage operated in the linear region shouldn't be much different than a good solid state preamp aside from a little more Johnson noise.