1. Home
  2. Electronics Design
  3. Low Noise Direct Coupled Preamp

Low Noise Direct Coupled Preamp

A number of designs exist of preamplifiers to measure power supply, zener, LED, and other noise sources. These generally have a large input cap to isolate the source voltage from the input of the preamp. A large capacitor can have huge leakage, which can disrupt the input bias of the preamp.

Here is a proof-of-concept approach that doesn't use a large series input cap. Instead, the preamp input is direct coupled to the source. It uses a LT1028, which has an unparalleled low frequency noise corner and very low input noise.

The gain is set to 40dB. The low frequency response is determined by large electrolytics in the feedback path, but the leakage current of the capacitors has minimal effect on the output due to the design.

As in other approaches, multiple preamps could be paralled for very low input noise. The feedback resistors could be adjusted for minimum overall noise.

You can download the LTspice ASC and PLT files here:

formatting link 

--
The best designs occur in the theta state. - sw
Reply to
Steve Wilson
Loading thread data ...

Steve Wilson wrote: =========================

** Horse manure.

A 1uF film cap ( with SFA leakage) feeding a 10Mohm load gives corner frequency of 15 mHz.

If you have a point, you did not post it.

..... Phil

Reply to
Phil Allison

Nah, trying to do this with AC coupling is useless in the sub-audio range.

The tempco of polypropylene is about +200 ppm/K. With even 1V DC on it, that cap will drift -0.2 mV/K. To get down to the ~1 nV/sqrt(Hz) level, the temperature fluctuations have to be below 5 uK/sqrt(Hz). Pretty tough at low frequency.

And then there's the drift, which is even worse.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com
Reply to
Phil Hobbs

Yes, the LT1028 is very low noise at audio-ish frequencies but it has a non-trivial noise peak around (from memory) 6MHz that is conveniently not shown on the datasheet, or at least wasn't back in the day. I don't know if the noise peaking is in the LTSPICE model. This may or may not matter, depending on your measurement bandwidth and how you set it.

If you want the same noise performance but usable to MHz, Analog's AD797 is better behaved with no noise bump. The AD797's designer showed me spectrum analyzer plots way back when.

Reply to
Steve Goldstein

That has the same temperature coefficient problem as Phil A.'s suggestion, only much worse on account of the electros. (Sorry, Phil!)

You're much better off with a MCU-adjusted voltage divider (using some DAC with a thin-film resistor network) zeroing out the DC at the beginning of the measurement.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com
Reply to
Phil Hobbs

The AD797 does not match the 1/f noise of the lt1028. The typical lt1028 corner frequency is 3.5 Hz. The AD797 noise plot stops at 10 Hz.

The LT1028 noise bump is 5nV/root(Hz) at 400KHz. This is not a large peak. The preamp is designed to roll off before then.

--
The best designs occur in the theta state. - sw
Reply to
Steve Wilson

Now that we have a way to measure power supply noise, we need a power supply with very low ripple, especially at DC. This one provides the following ripple attenuations:

DC : -50dB 1 Hz : -90 dB 10 Hz : -110 dB 100 Hz : -130 dB 1 KHz : -154 dB 10 KHz : -170 dB 100 KHz : -180 dB 1 MHz : -160 dB

Having the low ripple supply and low noise preamp means we can ping-pong back and forth to improve both designs.

You can download the Low Ripple supply here:

formatting link 

--
The best designs occur in the theta state. - sw
Reply to
Steve Wilson

========================

** Crapology.

The OP only spoke of cap leakage - f*****ad. --------------------------------------------------------------------------

** Likely never matter in real applications involving DC PSUs, zeners and the like.

Wot a colossal wanker you are.

........... Phil

Reply to
Phil Allison

Who wants 1 nV below a hertz or so? ,I once had this idea to measure 1/f noise from DC, so I made a bridge (All 1 k ohm 0.1% MF resistors to start* IIRC) And all I remember was DC drifting one way or the other.. (Thermal effects maybe) I didn't spend much time on it.

George H.

  • replace one with carbon composite later.
Reply to
George Herold

is very low. Any low noise measurements require the circuits be placed in a shielded box. This minimizes air currents that contribute to low frequency drift. The preamp response starts rolling off around 0.2Hz, so long-term drift is not measured.

The temperature coefficient and leakage of the electros has negligible effect on the output voltage due to their placement in the feedback path.

You can delete one of the paired electrolytics, and change the leakage current with negligible effect onthe output voltage.

I don't think you understand the purpose of this circuit. It is a low noise preamplifier with a gain of 40 dB.

A voltage divider has no amplifier. You need to include one.

The usual approach is to add a series coupling capacitor at the input of a low noise amplifier. The cap can have significant leakage, which affects the bias of the preamp.

An example is the 70 pV/root(Hz) preamp in AOE. It specifies a *large* input cap. My preamp has no input cap.

--
The best designs occur in the theta state. - sw
Reply to
Steve Wilson

Once you put a charge on the cap, its tempco turns into a voltage drift.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com
Reply to
Phil Hobbs

At frequencies for which the thermal mass approximation still works, the amplitude of thermally-forced noise and drift goes as 1/f. That extends the sensitivity up to surprisingly high frequencies, especially in a powered test circuit with people moving around.

Making good measurements at sub-hertz frequencies is surprisingly hard.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com
Reply to
Phil Hobbs

It is more expensive, and depends on a much more elaborate manufacturing process, but it does seem to be a wonderful part.

I earned a few brownie points once by recommending it as an alternative to the LT1028 (which I had used in several designs that had made it into production) purely on the basis of the data sheet. It did turn out to do a much better job in the application.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

When you're talking noise, you always have to keep in mind the application and its requirements.

The limit can also be in the sensor, or the source, never mind the amplifier, output transducer or recorder.

Noise is actually a borrowed term applied to any random background influence in the quantity of interest. Separation or cancelation techniques are often more important than 'gain'.

RL

Reply to
legg
[...]

Q = CV, so V = Q/C. Temperature changes C, so V changes with the same Q.

The measurement must be made in a shielded box to eliminate external RFI.

The box also eliminates fluctuating air currents that could change the temperature of the electrolytics.

Any drift due to temperature changes will be very low frequency. The gain of the preamplifier starts to roll off at 0.2 Hz, so any drift due to temperature change will be strongly attenuated.

If necessary, the preamplifier could also be immersed in mineral oil, further reducing any temperature changes.

Furthermore, the main intent of the preamp is to measure frequencies above

1 Hz. Flicker noise will probably be severe at 1 Hz and below, and will probably outweigh any drift due to temperature coefficient.

Furthermore, the low frequency rolloff could be raised by reducing the capacitance at the negative input of the op amp. The low frequency and high frequency rolloff could be adjusted by switching in different capacitors to control the bandwidth of the preamp.

As stated, this is a proof-of-concept approach to eliminate the coupling capacitor at the input of the preamp. It is not a finished design ready for production. Further refinement can be made to enhance the performance and reduce any errors, but you have to consider the cost of refinement vs the perfrmance expected, and decide if the performance is already good enough.

In this case, I think flicker noise will be the limiting factor at low frequencies. Capacitor temperature coefficient is an interesting concept, but I tend to think it will be buried in flicker noise.

--
The best designs occur in the theta state. - sw
Reply to
Steve Wilson

I've used Phil's technique, a 1 uF film cap and big R. to measure power supply noise. It worked fine... (I still remember being surprised the big R didn't add to the noise, mostly dominated by the opamp in this case.)

Grin, for the above case if I tried to measure the power supply noise below ~100 Hz, at maximum gain, it would start to oscillate as soon as I looked at it sideways.

George H.

Reply to
George Herold

George Herold wrote: =================

** If the noise coming from your DC PSU is dominated by the input noise voltage of a modern BiFET op-amp ( 16nV/rtHz ) .... you need bigger things to worry about.

..... Phil

Reply to
Phil Allison

And you forgot that you get the full broadside of the 10 Meg resistor voltage noise at 15 mHz. A huge input capacitor is needed to short that noise through the low impedance DUT at these low frequencies.

That is not welcome because the settling time goes to infinity, leading to a not so satisfying user experience; especially since the settling time restarts for every input DC change.

If you see a noise corner at 50 Hz that goes up at least 20 dB/decade (1/f is 10 dB/decade), then you are a victim of the small input capacitor.

Any expectations of usability at 15 mHz are void then, of course.

Gerhard

Reply to
Gerhard Hoffmann

Oh in this case the supply was to bias other noise measurements. Powering leds or light bulbs for shot noise measurements. or just resistors... Mostly with 8nV/rtHz opa134's, with some averaging you can measure ~1 nV. (you need a switch to ground the front end and measure the noise w/o the supply.) George H.

Reply to
George Herold

Is charge actually conserved in (some sort of) capacitor when the temperature changes?

Is energy conserved?

If you pull the plates of a parallel-plate vacuum cap apart, charge is conserved but energy is not.

--

John Larkin         Highland Technology, Inc   trk 

The cork popped merrily, and Lord Peter rose to his feet.   
"Bunter", he said, "I give you a toast. The triumph of Instinct over Reason"
Reply to
John Larkin

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.