Pre-amp Circuit Help

This is the place for any magazine-related discussions that don't fit in any of the column discussion boards below.
andyroy15
Posts: 5
Joined: Tue Mar 02, 2004 1:01 am
Location: UK
Contact:

Pre-amp Circuit Help

Post by andyroy15 »

I am currently trying to build a 40dB pre-amp for an array of acoustic transducers receiving 0.5ms burst signals at 70KHz, and need a bit of help.<p>The complication is that it needs to be fairly small and portable, and run on a -9 to +9 volt battery supply. The current circuit of a series of 741's is not adequate enough at the frequency.<p>Luckily I am only trying to detect the start of the signal (a sort of 1-bit ADC) and so I'm not hugely worried about a bit of distortion within it, as long as the signal is readable.<p>Can anyone help or point me in the right direction? Thanks.<p>[ March 02, 2004: Message edited by: andy15 ]</p>
cato
Posts: 366
Joined: Sun Sep 07, 2003 1:01 am
Contact:

Re: Pre-amp Circuit Help

Post by cato »

It sounds like you understand the problem, but since you asked the question, I assume you are unaware of the name of the specification you need to be concerned about. You need to shop for an amplifier with a better GAIN-BANDWIDTH PRODUCT. I'm too groggy to convert 40db into a straight multiplication factor. I'll leave that for you to figure out. When you figure that out, multiply it by the 70Khz. That will tell you the gain bandwidth product you need. Shop for an op amp that will provide that level of gain-bandwidth. Alternatively, cascade a series of op amp circuits, never exceeding the gain-bandwith product rating of each op amp (use low enough gain in each stage so the gain times 70Khz is less than the GBP of the device.<p>Good luck
User avatar
haklesup
Posts: 3136
Joined: Thu Aug 01, 2002 1:01 am
Location: San Jose CA
Contact:

Re: Pre-amp Circuit Help

Post by haklesup »

I agree with cato in that the issue may not be the circuit design but the selection of the component.<p>The 741 is an old design and has very loose specs. Presently, there are many op amps with excellent specs. Type "op amp selection guide" into a search engine and go from there. It is likly that you will be able to find a circuit to copy in one of the application notes that often accompany the data sheets for these devices. At 70Khz, circuit construction is not so critical but keep everything neat and short and it should work out.<p>National Semiconductor and TI have extensive selections. Don't forget to search the app notes at the Maxim IC site for ideas as well. NS has a cool online design tool that recommends parts based on the specs of the circuit you want to design Amp designer
User avatar
Chris Smith
Posts: 4325
Joined: Tue Dec 04, 2001 1:01 am
Location: Bieber Ca.

Re: Pre-amp Circuit Help

Post by Chris Smith »

They make nice FET replacments for the 741.
rshayes
Posts: 1286
Joined: Tue Mar 04, 2003 1:01 am
Contact:

Re: Pre-amp Circuit Help

Post by rshayes »

A gain of 40 db corresponds to a voltage gain of 100. For a bandwidth of 70 KHz, the minimum gain bandwidth of the op amp should be over 7 MHz.<p>The 741 class of op amps usually had a gain bandwidth product well below 1 MHz, and would be unsuitable for this application.<p>Analog Devices and Comlinear (now part of National Semiconductor) made op amps that would meet this requirement. Some of the faster TL types made by Texas Instruments might also be suitable.<p>Power supply bypassing is important, since these devices still have gain in the 10 MHz region. Ceramic capacitors (usually .1 uF) located close to the power pins of the op amp are a good starting point, with large electrolytic capacitors (10 to 100 uF) located further away.<p>If only a few channels are involved, consider using two stages with a gain of 10 in each stage. Then the gain bandwidth required becomes 700 KHz, and the 741 type of amplifier may be usable (just barely).<p>Watch out for slew rate. A 3.5 volt RMS sine wave at 70 KHz has a slew rate of 2.2 volts per microsecond. This is about 4 or 5 times faster than a 741 can handle. More recent amplifiers than the 741 are considerably better in this respect.
andyroy15
Posts: 5
Joined: Tue Mar 02, 2004 1:01 am
Location: UK
Contact:

Re: Pre-amp Circuit Help

Post by andyroy15 »

Sorry, should of clarified the gain, its 40dB in terms of signal, so I need a gain of at least 10000.<p>Thanks for the help
cato
Posts: 366
Joined: Sun Sep 07, 2003 1:01 am
Contact:

Re: Pre-amp Circuit Help

Post by cato »

Ok, I've had a good nights sleep AND a mug and a half of strong Maxwell House Columbian Supream. However, it has been 20 years since I've worried about such things.... Therefore I'm only pretty sure that the formula for converting voltage gain to dB is:<p>20 log Vin/Vout = dB<p>The log of 100 is 2. Therefore I have to agree with Stephen. 40db is a gain of 100. Of course there is the issue of Power gain...I think the formula for that is:<p>10 log Pin/Pout = dB<p>...which relates 40dB to 10000....<p>Of course...no amount of coffee will help me remember if I have the formulas reversed...which I guess is what andy15 is suggesting....<p>[ March 05, 2004: Message edited by: cato ]</p>
andyroy15
Posts: 5
Joined: Tue Mar 02, 2004 1:01 am
Location: UK
Contact:

Re: Pre-amp Circuit Help

Post by andyroy15 »

Again, sorry, I need a gain of at least 10000. So in voltage terms it is 80dB. It's also been a few years for me too!<p>Would there be any side effects to using 2 40dB amps to multiply up to the total 80dB?<p>[ March 03, 2004: Message edited by: andy15 ]</p>
cato
Posts: 366
Joined: Sun Sep 07, 2003 1:01 am
Contact:

Re: Pre-amp Circuit Help

Post by cato »

In an ideal world there should be no side effects. Infact, that is what I was suggesting... or 4 20dB amps...or 8 10dB amps...The only issues I see are those having to do with offset voltages getting amplified, but ac coupling can deal with that, and added noise...but in your application that shouldn't matter too much....<p>You might consider using one or more instrumentation amps (three opamps in a particular arangement) Analog Devices and Burr Brown used to make these is a single package...I forget WHY you might consider using an instrumentation amp....but I do think they might be used in this application.
rshayes
Posts: 1286
Joined: Tue Mar 04, 2003 1:01 am
Contact:

Re: Pre-amp Circuit Help

Post by rshayes »

The purpose of an instrumentation amp is to amplify signals that are referenced to a point other than the local ground. They are usually arranged for very high rejection of signals common to both inputs. This is normally only necessary for the input stage of an amplifier.<p>With a gain of 10000, the input voltage is in the 100 microvolt range. This amount of gain will probably require several stages, probably three or four. Shielding will probably be necessary on the input. The input stage will have to be designed to match the transducer, and should have a low noise level. The later stages can be op amps in conventional configurations.<p>Ideally, the input stage should be physically close to the transducer. Every inch of wire at this point is an opportunity for noise pickup.<p>The 100 microvolt level is getting close to the point where thermal noise is significant. The thermal noise in a 1 megohm resistor is about 35 microvolts in a 70 KHz bandwidth. Lower values of resistance generate lower values of noise. Carbon or carbon film resistors should be avoided in the first stage, since they will contribute additional noise.<p>The power source for the first stage or two will need additional filtering and decoupling. Most electrolytic capacitors, such as would be used as a power supply bypass (in addition to the local ceramic bypasses) have an Equivalent Series Resistance (ESR) in the .1 ohm region. If the output signal causes a change in power supply current of 1 milliamp, the change in power supply voltage will be 100 microvolts. If this voltage change is coupled to the input, a feedback loop with an overall gain greater than unity may be formed, and the circuit will oscillate.<p>Some op amps may have low enough noise characteristics for the input stage. Bipolar types tend to have lower voltage noise. These are useful for low impedance sources, where current noise is less significant. Higher impedance transducers may require FET types, since the curent noise becomes very significant.<p>A discussion of noise in operational amplifiers was published in an early edition of "Analog Dialog", specifically the March 1969 issue. "Analog Dialog" is published by Analog Devices. This article is available in PDF form from the Analog Devices web site (www.analog.com) as part of "The Best of Analog Dialog- 1967 to 1991". Look in the archive section of "Analog Dialog".
dyarker
Posts: 1917
Joined: Fri Aug 22, 2003 1:01 am
Location: Izmir, Turkiye; from Rochester, NY
Contact:

Re: Pre-amp Circuit Help

Post by dyarker »

40dB per stage sounds like an oscillator unless you're REALLY careful with layout. I'd think 10dB very low noise section, two 30dB sections, and a 10dB last section.<p>[ March 04, 2004: Message edited by: Dale Y ]</p>
Dale Y
cato
Posts: 366
Joined: Sun Sep 07, 2003 1:01 am
Contact:

Re: Pre-amp Circuit Help

Post by cato »

Yeah....what they said....that's what I meant.
cato
Posts: 366
Joined: Sun Sep 07, 2003 1:01 am
Contact:

Re: Pre-amp Circuit Help

Post by cato »

oooo ooooo ooooo I think I see a way to simplify your project....If I understand the first post, all you're trying to do is detect the leading edge of an envelope of 35 cycles (.5 milliseconds wide). That moves us from the ritzy 70Khz neighborhood to the cheaper 2Khz side of the tracks. Some lowpass filtering up front (with a cut off of say 4Khz) should reduce your GBW problems considerably.<p>Steve, Dale -waddaya think?
Ron H
Posts: 360
Joined: Tue Dec 18, 2001 1:01 am
Location: Boise, ID
Contact:

Re: Pre-amp Circuit Help

Post by Ron H »

<blockquote><font size="1" face="Verdana, Helvetica, sans-serif">quote:</font><hr>Originally posted by cato:
oooo ooooo ooooo I think I see a way to simplify your project....If I understand the first post, all you're trying to do is detect the leading edge of an envelope of 35 cycles (.5 milliseconds wide). That moves us from the ritzy 70Khz neighborhood to the cheaper 2Khz side of the tracks. Some lowpass filtering up front (with a cut off of say 4Khz) should reduce your GBW problems considerably.<p>Steve, Dale -waddaya think?<hr></blockquote>
I'm not Steve or Dale, but I have an opinion anyway. ;) The received signal probably won't have any energy below 4kHz, or even below 70kHz. You are probably thinking of a unipolar or raised sine burst, but I'm guessing the received signal will be symmetrical about the baseline.<p>[ March 05, 2004: Message edited by: RonH ]</p>
rshayes
Posts: 1286
Joined: Tue Mar 04, 2003 1:01 am
Contact:

Re: Pre-amp Circuit Help

Post by rshayes »

Unfortunately, you would first have to detect the envelope. If the amplitude of the signal is in the 100 microvolt range, you will be lucky to get a few microvolts out of a diode detector, since this will be in the region where the detector response is square law. The bandwidth would be lower, but the gain requirement would increase to a million or so. The output of the rectifier would probably disappear into the thermal noise of the amplifier. The thermal noise in a 30K resistor would be about a microvolt with a 2 KHz bandwidth. The noise density of the input stage would have to be well under 20 nanovolts per root hertz as well.<p>Overall, amplifying the signal before rectifying it might be the easier solution. A compromise might be to detect the signal at a level of about 100 millivolts or so.
Post Reply

Who is online

Users browsing this forum: Ahrefs [Bot] and 29 guests