Advertisement

Blog

When Is a Reference Not Adequate? It Depends

In the design of analog front ends that require low noise, it is important that all aspects of the design are considered. This is especially true when looking at design targets with 20+ bits of resolution for, say, power metering or other low-frequency audio applications. When measuring a circuit capable of producing such high performance, creating a board with the proper references is a critical component of success.

The design of an evaluation board to properly measure a part with high dynamic range and low noise requires that the board include a low noise reference. But how low is low? To address this question, I will present a case where we experienced degraded performance due to issues with reference noise. To explain the issue, I will provide some background information on sampling analog-to-digital converters.

For a sampling ADC, the reference voltage supplied to the part determines the ultimate quality of the converted signal. In most ADCs used for general-purpose applications in the range of 10 bits of resolution, creating a board reference is not too difficult if attention to detail and proper bypassing rules are followed. To illustrate this point, Figure 1 shows a typical reference that a board design may use to develop the voltage the ADC will use for conversion. As can be seen in the picture, the board is designed with regulators that can tolerate large decoupling capacitors. In addition, the decoupling caps are placed on the supplies to the regulators.

Figure 1

Typical reference used for board design.

Typical reference used for board design.

Notice, if you will, that the bypass caps are made up of a series of smaller caps in parallel, with at least one of those smaller caps placed as close to the DUT as possible. For more information on this topic, refer to my previous blog. Once these caps are properly placed on the board and are targeted for adequate filtering, is the reference design adequate? It depends. What about the low-frequency noise coming out of the regulated voltage used for the reference?

To answer these questions, I must talk briefly about what happens to noise that exists on the golden reference voltage of the ADC. The reference and its noise is sampled just as if it were another input, as shown in Figure 2 for a typical input sampling stage of an ADC. Therefore, any noise or unwanted energy is sampled along with the signal. The noise sampling is more or less of a problem depending on the architecture chosen. Nevertheless, the existence of noise is problematic for very high-performance ADCs with 21+ bits of resolution and bandwidths that process signal on the DC+ to audio bandwidths, as is the case for power metering front ends mentioned in this example.

Figure 2

Typical sampling ADC stage.

Typical sampling ADC stage.

The data shown in Figure 3 is taken from a 21-bit analog front end design for exactly this market. The ADC that was part of the analog front end was a sigma delta ADC that used the regulated supply as the reference for the ADC. Extreme attention to detail was paid when designing the board. This detail included the choice of regulators for the references, as well as the placement of capacitors to filter the supply. In addition, special attention was given to the routing of the traces on the board. The results shown in Figure 3 explain what happened.

Figure 3

Plotted performance of the sampling AFE.

Plotted performance of the sampling AFE.

The plot shown provides the FFT of the output of this front end before and after a solution was provided to fix the problem. The trace in blue on the plot shows a strange noise shelf around the signal process. This noise shelf accounts for ∼10 dB in loss performance. Why? To answer this question, remember my previous statement about the sampling stage shown in Figure 2. In the case of the integrator sampling stage of the sigma delta ADC shown in Figure 2, the integrator design used the reference as another input that is sampled just like the signal. Therefore, any noise on the reference is sampled just like the signal. As can be seen in the plot, there is low-frequency noise on the reference that is sampled and shows up in the FFT. To combat this problem, I had to design a 0.5Hz filter into the regulated reference to reduce the noise before it was sampled and folded into converted bandwidth (shown in Figure 1).

In most cases, regulated noise is usually not a problem unless these voltages are used as references for high-performance ADCs. In this case, every aspect of the board design is not only important but crucial.

Have you experienced similar situations where the unexpected noise from a reference caused issues?

48 comments on “When Is a Reference Not Adequate? It Depends

  1. green_is_now
    October 23, 2014

    If you have higher voltage linear regulators in your system, even if they are multiuse LR's, If enough headroom is present, then cascade into a dedicated LR just for the ADC.

    Any capacitor has a finite Rs in series with capacitance. This when exited by high frequency cause HF differential mode noise out of common mode noise by the ratio of input impedance to capacitor impedance.

    Noticed your red paint brush covers up the blue hash at high frequency.

    This is another example of the squishyness of analog solutions.

    Without attacking the source for reduction you can only push the noise around in the FFT!

    But this is still valid because of better relative performance trade offs of the high and low frequency penalties in the particular system.

    if you know the source it aint noise!

    Thus you can fix it with active amplitude cancelation upstream of LR and filter if it is trully low frequency problem AND you can create a signal from a node in your system.

    This is not possible unless it is relativey slow relative to feedback delay skew AND you can recreate the signal without curruption (noise ) also encroaching into signal.

    or fix/reduce the issue at the source AKA SMPS.

  2. Myled
    October 24, 2014

    “The design of an evaluation board to properly measure a part with high dynamic range and low noise requires that the board include a low noise reference. But how low is low? To address this question, I will present a case where we experienced degraded performance due to issues with reference noise.”

    Brandit, “how low is low” is a good and confusing question because noise acceptances level various from board to board and from circuit to circuit. In certain circuits noise acceptance level is almost zero because any noise can degrade the performance.

  3. vasanjk
    October 25, 2014

    Brandt, Very informative post. I would like to know if it would help to introduce a unity gain buffer after an external reference source. Also, would there be a performance degradation in terms of noise as we are introducing an additional component between ADC and reference source. In essence, which is better, the combination or a specific reference ic which does not need a buffer?

  4. geek
    October 26, 2014

    “because noise acceptances level various from board to board and from circuit to circuit. In certain circuits noise acceptance level is almost zero because any noise can degrade the performance.”

    @Mnaynalog: That's a valid concern. I think not only it depends on the circuit but also on the application of the circuit itself. The noise level for the same circuit in a different application can be higher or lower than in another application. That also needs to be considered.

  5. geek
    October 26, 2014

    “In most cases, regulated noise is usually not a problem unless these voltages are used as references for high-performance ADCs. In this case, every aspect of the board design is not only important but crucial.”

    @Brandt: What are the potential problems that can arise if noise isn't seriously taken care of? Can the damages be really strong enough to render the application useless?

  6. dassa.an
    October 27, 2014

    @TZUBAIR: Well you have to take care and control the noise. If not it will be very damaging and the risk it involves is really high. 

  7. dassa.an
    October 27, 2014

    @tzubair: Do we have to take the average and apply it or assume based on previous records ? Which one is more accurate ? 

  8. PCR
    October 27, 2014

    dassa.an I feels that the  based on previous record will more accurate isn't it ?

  9. goafrit2
    October 27, 2014

    >>  I would like to know if it would help to introduce a unity gain buffer after an external reference source

    Always a good strategy – that gives you a low impedance node to drive bigger loads. Yet that does not mean a noisy reference is off the hook

  10. goafrit2
    October 27, 2014

    >> In certain circuits noise acceptance level is almost zero because any noise can degrade the performance.”

    In most biomedical circuits that is the case since the signals are very small – uV or most mV – so any noise the circuit is dead. Reference designs are challenging. Tobi Delbruch of Swiss technical institute has sone great ideas on how to make references for challenging small signals.

  11. fasmicro
    October 27, 2014

      What are the potential problems that can arise if noise isn't seriously taken care of?

    When the signal is corrupted, you simply do not have a product anymore. This is a big issue for AFE in MEMS and circuit interface

  12. fasmicro
    October 27, 2014

    In this case, every aspect of the board design is not only important but crucial

    Why board design gives you the edge to waste components and get over issues, ASIC design is where reference becomes a big problem. When that reference begins to supply all parts of the circuits signal, if it has not been well designed, the system cannot optimally perform. I do not do a lot of boards, I make ASICs. Refernce design is a task therein.

  13. vasanjk
    October 27, 2014

    @goafrit Noisy references do not become quiet when a buffer is introduced. The noise too takes advantage of the benefits of a buffer. The best way forward would be any of the following and more as in the blog: Positioning ref source v near adc, using proper caps as per mfr guidelines, proper pcb layout and so on.

  14. chirshadblog
    October 28, 2014

    @vasanjk: Isn't there a way of gaining some valuable output from the noise itself ? If so we can make it more productive. 

  15. chirshadblog
    October 28, 2014

    @Ranasinghe & Dass.an: How do you'll assume it ? Any experience on this or just an assumption ? 

  16. vasanjk
    October 28, 2014

    @chirshad, Noise can be productively utilized, but not here. When you do oversampling and averaging, you could squeeze out a couple of more bit resolution from an ADC, with the help of the noise present in the signal.

  17. geek
    October 28, 2014

    “In most biomedical circuits that is the case since the signals are very small – uV or most mV – so any noise the circuit is dead”

    @goafrit2: I agree. I think the medical appliances might be the most critical type of electronic devices where the signal accuracy needs to be very accurate. Any noise, as you said, will render the circuit useless and may result in a huge malfunction of the overall appliance.

  18. geek
    October 28, 2014

    “When the signal is corrupted, you simply do not have a product anymore. This is a big issue for AFE in MEMS and circuit interface”

    @fasmicro: Do you know about any tools that can be used to test the accuracy of signals at any point in time in a circuit and the level of noise that may be occurring. I'm talking about tools that can be remotely used to monitor and don't have to be installed within the circuit itself.

  19. Myled
    October 29, 2014

    “That's a valid concern. I think not only it depends on the circuit but also on the application of the circuit itself. The noise level for the same circuit in a different application can be higher or lower than in another application. That also needs to be considered.”

    Tzubair, by circuit I meant the entire board. Sometime improper earthing can also cause noises.

  20. Myled
    October 29, 2014

    “Noise can be productively utilized, but not here. When you do oversampling and averaging, you could squeeze out a couple of more bit resolution from an ADC, with the help of the noise present in the signal.”

    Vasanjk, how we can productively utilize these noises and how we can tap it. So far I hadn't see any such applications.

  21. vasanjk
    October 29, 2014

    @Myanalog, I Noise is something that is unreliable and I don't think anybody has used noise that effectively. However, if you are using a 10 bit adc and would like one or two bits more in resolution, the normal practice is to use oversampling and averaging to achieve that. But this technique can work only with some noise coupled to the signal. As far as I know, this is the only place, noise helps. ( you can correct me if I am wrong.)

  22. chirshadblog
    October 30, 2014

    @vasanjk: Isnt there any other way of reducing this noise ? 

  23. chirshadblog
    October 30, 2014

    @Myanalog: Yes if there are any possibilities, I guess there will be solutions for many of our issues related to noise. 

  24. Sachin
    October 31, 2014

    When you do oversampling and averaging, you could squeeze out a couple of more bit resolution from an ADC, with the help of the noise present in the signal.

    @vasanjk, are you referring to SNR ratio ? I thought we always expected noise level to remain as much low as possible to gain more SNR. I am curious to know how noise helps us to squeeze out more bit from ADC.

  25. Sachin
    October 31, 2014

    Always a good strategy – that gives you a low impedance node to drive bigger loads. Yet that does not mean a noisy reference is off the hook

    @goafrit2, I am curious to know what is the solution for nosiy reference ? How can we make sure that nosiy reference doesnt impact the perfromance of the circuit ?

  26. Sachin
    October 31, 2014

    What are the potential problems that can arise if noise isn't seriously taken care of?

    @tzubair, If noise is present in the circuit then it reduces the accuracy of the circuit. For example if noise is present in ADC then it would reduce the number of resolution bits which means we can recover less data.

  27. Sachin
    October 31, 2014

    Brandit, “how low is low” is a good and confusing question because noise acceptances level various from board to board and from circuit to circuit. 

    @Myanalog, there is no straight answer to this. I think it varies. Some application can tolerate higher level of noise and some dont. So you need to set the noise level based on your application.

  28. dassa.an
    October 31, 2014

    @sachinee: True you do need to keep an eye on the noise levels. Balancing them would be the difficult part but still its vital. 

  29. dassa.an
    October 31, 2014

    @Sachinee: So basically the noise levels have to be balanced properly to gain more valuable output. 

  30. dassa.an
    October 31, 2014

    @SachinEE: Well a good question indeed. I think there are ways but has anyone tried it ? 

  31. dassa.an
    October 31, 2014

    @SachinEE: Well there are many benefits in it. It can be used for greater things

  32. fasmicro
    November 1, 2014

    >> Noisy references do not become quiet when a buffer is introduced. The noise too takes advantage of the benefits of a buffer

    Absolutely, the best strategy is circuit is not processing noise. You simply need to avoid picking noise to start with. 

  33. fasmicro
    November 1, 2014

    >> Isn't there a way of gaining some valuable output from the noise itself 

    I do not understand the question. Do you mean extracting signal from noise? You can do that if the noise has indeed signal. In ADC design, that is allowed where you introduce noise to help get more bits but that is not refenece noise though!

  34. fasmicro
    November 1, 2014

    >> When you do oversampling and averaging, you could squeeze out a couple of more bit resolution from an ADC, with the help of the noise present in the signal.

    Sure but that is not REFERENCE noise. Oversampling and averaging will require special noise sources being introduced in the circuit. You do not use reference for that.

  35. fasmicro
    November 1, 2014

    >> Do you know about any tools that can be used to test the accuracy of signals at any point in time in a circuit and the level of noise that may be occurring. 

    A good probe station should have capacity to give you some frequency components. However, there are different types of noise in circuit and that could render any strategy useless.

  36. fasmicro
    November 1, 2014

    “That's a valid concern. I think not only it depends on the circuit but also on the application of the circuit itself. The noise level for the same circuit in a different application can be higher or lower than in another application. That also needs to be considered.”

    That is what drives the cost of products. If you are making a toy, the noise tolerance will be different to when you are making a medical device. Noise is relative, it all depends on the product and applications as you noted.

  37. goafrit2
    November 1, 2014

    how we can productively utilize these noises and how we can tap it. So far I hadn't see any such applications.

    In ADC, it is very common. You can get extra bits by adding noise to your signal. 

  38. goafrit2
    November 1, 2014

    Well a good question indeed. I think there are ways but has anyone tried it ? 

    Oversampling ADCs use that all the time. It is nothing new. The goal is to get more resolutions.  However, be informed that some of those extra bits are marketing bits. The DNL and INL may show that the bits may not be practically useful

  39. goafrit2
    November 1, 2014

    The most challenging circuits I have worked with respect to noise are in biomedical systems and MEMS analog front ends. These are areas where any small noise budget makes the system useless because the signals are very small. But the good news is that when you are working on biomedical designs, the products will be costing more and that means you can do the right things to reduce noise from cropping up

  40. vasanjk
    November 2, 2014

    goafeit2 “But the good news is that when you are working on biomedical designs, the products will be costing more and that means you can do the right things to reduce noise from cropping up” Not any more. Chinese products are flooding the market with low end prices while having outputs 'resembling' the brands. Longevity and robustness are a question, though. Still they force the prices of indigenous products low due to competition. Coming to noise, a typical ecg system requires at least 3 notch filters and 1 band pass filter to keep noises away.the biggest culprit is the mains hum.

  41. fasmicro
    November 3, 2014

    >>  Chinese products are flooding the market with low end prices while having outputs 'resembling' the brands.

    In biomedical, even for Chinese products, they are largely more expensive to the overall consumer sector. Also, I do not know doctors which are quickly using Chinese biomed products.

  42. vasanjk
    November 3, 2014

    fasmicro A typical ECG recorder costs $250 from a Chinese manufacturer, while a simple weighing scale costs as much.A local manufacturer of ECG recorder is forced to keep his prices down. Many physicians in India use such products in their clinics. I agree hospitals may not do so.

  43. Myled
    November 5, 2014

    “Yes if there are any possibilities, I guess there will be solutions for many of our issues related to noise. “

    Chirsh, yes there are some noise reduction or filtration methods, which can be used with boards.

  44. Myled
    November 5, 2014

    ”  Noise is something that is unreliable and I don't think anybody has used noise that effectively. However, if you are using a 10 bit adc and would like one or two bits more in resolution, the normal practice is to use oversampling and averaging to achieve that. But this technique can work only with some noise coupled to the signal. As far as I know, this is the only place, noise helps. ( you can correct me if I am wrong.)”

    Vasan, thanks for the clarification. So far I hadn't used noises effectively; just filtered out.

  45. Myled
    November 11, 2014

    “In ADC, it is very common. You can get extra bits by adding noise to your signal. “

    Goafrit2, I understood it from vasan's comment.

  46. Myled
    November 11, 2014

    “That is what drives the cost of products. If you are making a toy, the noise tolerance will be different to when you are making a medical device. Noise is relative, it all depends on the product and applications as you noted.”

    Fasmicro, you are right and I mentioned exactly the same.

  47. Myled
    November 11, 2014

    “there is no straight answer to this. I think it varies. Some application can tolerate higher level of noise and some dont. So you need to set the noise level based on your application.”

    SachinEE, you are right. The noise acceptance level can be varying from application to application.

  48. goafrit2
    December 6, 2014

    >> A typical ECG recorder costs $250 from a Chinese manufacturer, while a simple weighing scale costs as much.A local manufacturer of ECG recorder is forced to keep his prices down

    That is certainly in some developing parts of the world. But it comes to U.S., be looking for thousands of dollars. It has been expensive for U.S. hospitals to buy them!

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.